Comparison of stretched-Exponential and monoexponential model diffusion-Weighted imaging in prostate cancer and normal tissues

Prediction of Chemotherapy Response in Locally Advanced Breast Cancer Patients at Pre-Treatment Using CT Textural Features and Machine Learning: Comparison of Feature Selection Methods

RATIONALE

Neoadjuvant chemotherapy (NAC) is a key element of treatment for locally advanced breast cancer (LABC). Predicting the response of NAC for patients with LABC before initiating treatment would be valuable to customize therapies and ensure the delivery of effective care.

OBJECTIVE

Our objective was to develop predictive measures of tumor response to NAC prior to starting for LABC using machine learning and textural computed tomography (CT) features in different level of frequencies.

MATERIALS AND METHODS

A total of 851 textural biomarkers were determined from CT images and their wavelet coefficients for 117 patients with LABC to evaluate the response to NAC. A machine learning pipeline was designed to classify response to NAC treatment for patients with LABC. For training predictive models, three models including all features (wavelet and original image features), only wavelet and only original-image features were considered. We determined features from CT images in different level of frequencies using wavelet transform. Additionally, we conducted a comparison of feature selection methods including mRMR, Relief, Rref QR decomposition, nonnegative matrix factorization and perturbation theory feature selection techniques.

RESULTS

Of the 117 patients with LABC evaluated, 82 (70%) had clinical-pathological response to chemotherapy and 35 (30%) had no response to chemotherapy. The best performance for hold-out data splitting was obtained using the KNN classifier using the Top-5 features, which were obtained by mRMR, for all features (accuracy = 77%, specificity = 80%, sensitivity = 56%, and balanced-accuracy = 68%). Likewise, the best performance for leave-one-out data splitting could be obtained by the KNN classifier using the Top-5 features, which was obtained by mRMR, for all features (accuracy = 75%, specificity = 76%, sensitivity = 62%, and balanced-accuracy = 72%).

CONCLUSIONS

The combination of original textural features and wavelet features results in a greater predictive accuracy of NAC response for LABC patients. This predictive model can be utilized to predict treatment outcomes prior to starting, and clinicians can use it as a recommender system to modify treatment.

Histogram analysis of multiple mathematical diffusion-weighted imaging models for preoperative prediction of Ki-67 expression in hepatocellular carcinoma

Objective

To explore whether a combination of clinico-radiological factors and histogram parameters based on monoexponential, biexponential, and stretched exponential models derived from the whole-tumor volume on diffusion-weighted imaging (DWI) could predict Ki-67 expression in hepatocellular carcinoma(HCC).

Materials and Methods

Histogram parameters based on whole-tumor volumes were derived from monoexponential model, biexponential model, and stretched exponential model. Histogram parameters were compared between HCCs with high and low Ki-67 expression. Multivariate logistic regression and receiver operating characteristic curves were used to assess the ability to predict Ki-67 expression (expression index ≤ 20% vs. >20%).

Results

In the training and test set, the 5th percentile of distributed diffusion coefficient (DDC) yielded the area under the curve (AUC) value of 0.816 (95% CI 0.713 to 0.894) and 0.867 (95% CI 0.655 to 0.972), respectively. Multivariable analysis showed that alpha-fetoprotein (AFP) level, skewness of perfusion fraction(f), and 5th percentile of DDC were independent predictors of high Ki-67 expression in HCCs. In the training and test sets, the AUC of the combined model for predicting high Ki-67 expression in HCCs were 0.902 (95% CI 0.814 to 0.957) and 0.908 (95% CI 0.707 to 0.989), respectively.

Conclusion

Histogram parameters of multiple mathematical DWI models can be useful for predicting high Ki-67 expression in HCCs, and our combined model based on AFP level, skewness of f, and 5th percentile of DDC may be an effective approach for predicting Ki-67 expression in HCCs.

Apriori prediction of chemotherapy response in locally advanced breast cancer patients using CT imaging and deep learning: transformer versus transfer learning

Objective

Neoadjuvant chemotherapy (NAC) is a key element of treatment for locally advanced breast cancer (LABC). Predicting the response to NAC for patients with Locally Advanced Breast Cancer (LABC) before treatment initiation could be beneficial to optimize therapy, ensuring the administration of effective treatments. The objective of the work here was to develop a predictive model to predict tumor response to NAC for LABC using deep learning networks and computed tomography (CT).

Materials and methods

Several deep learning approaches were investigated including ViT transformer and VGG16, VGG19, ResNet-50, Res-Net-101, Res-Net-152, InceptionV3 and Xception transfer learning networks. These deep learning networks were applied on CT images to assess the response to NAC. Performance was evaluated based on balanced_accuracy, accuracy, sensitivity and specificity classification metrics. A ViT transformer was applied to utilize the attention mechanism in order to increase the weight of important part image which leads to better discrimination between classes.

Results

Amongst the 117 LABC patients studied, 82 (70%) had clinical-pathological response and 35 (30%) had no response to NAC. The ViT transformer obtained the best performance range (accuracy = 71 ± 3% to accuracy = 77 ± 4%, specificity = 86 ± 6% to specificity = 76 ± 3%, sensitivity = 56 ± 4% to sensitivity = 52 ± 4%, and balanced_accuracy=69 ± 3% to balanced_accuracy=69 ± 3%) depending on the split ratio of train-data and test-data. Xception network obtained the second best results (accuracy = 72 ± 4% to accuracy = 65 ± 4, specificity = 81 ± 6% to specificity = 73 ± 3%, sensitivity = 55 ± 4% to sensitivity = 52 ± 5%, and balanced_accuracy = 66 ± 5% to balanced_accuracy = 60 ± 4%). The worst results were obtained using VGG-16 transfer learning network.

Conclusion

Deep learning networks in conjunction with CT imaging are able to predict the tumor response to NAC for patients with LABC prior to start. A ViT transformer could obtain the best performance, which demonstrated the importance of attention mechanism.

A comparative study of mono-exponential and advanced diffusion-weighted imaging in differentiating stage IA endometrial carcinoma from benign endometrial lesions

PURPOSE

The purpose of the current investigation is to compare the efficacy of different diffusion models and diffusion kurtosis imaging (DKI) in differentiating stage IA endometrial carcinoma (IAEC) from benign endometrial lesions (BELs).

METHODS

Patients with IAEC, endometrial hyperplasia (EH), or a thickened endometrium confirmed between May 2016 and August 2022 were retrospectively enrolled. All of the patients underwent a preoperative pelvic magnetic resonance imaging (MRI) examination. The apparent diffusion coefficient (ADC) from the mono-exponential model, pure diffusion coefficient (D), pseudo-diffusion coefficient (D*), perfusion fraction (f) from the bi-exponential model, distributed diffusion coefficient (DDC), water molecular diffusion heterogeneity index from the stretched-exponential model, diffusion coefficient (Dk) and diffusion kurtosis (K) from the DKI model were calculated. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic efficiency.

RESULTS

A total of 90 patients with IAEC and 91 patients with BELs were enrolled. The values of ADC, D, DDC and Dk were significantly lower and D* and K were significantly higher in cases of IAEC (p < 0.05). Multivariate analysis showed that K was the only predictor. The area under the ROC curve of K was 0.864, significantly higher compared with the ADC (0.601), D (0.811), D* (0.638), DDC (0.743) and Dk (0.675). The sensitivity, specificity and accuracy of K were 78.89%, 85.71% and 80.66%, respectively.

CONCLUSION

Advanced diffusion-weighted imaging models have good performance for differentiating IAEC from EH and endometrial thickening. Among all of the diffusion parameters, K showed the best performance and was the only independent predictor. Diffusion kurtosis imaging was defined as the most valuable model in the current context.

Comparative study of stretched-exponential and kurtosis models of diffusion-weighted imaging in renal assessment to distinguish patients with primary aldosteronism from healthy controls

PURPOSE

To compare the ability of diffusion parameters obtained by stretched-exponential and kurtosis models of diffusion-weighted imaging (DWI) to distinguish between patients with primary aldosteronism (PA) and healthy controls (HCs) in renal assessment.

MATERIALS AND METHODS

A total of 44 participants (22 patients and 22 HCs) underwent renal MRI with an 11 b-value DWI sequence and a 3 b-value diffusion kurtosis imaging (DKI) sequence from June 2021 to April 2022. Binary logistic regression was used to construct regression models combining different diffusion parameters. Receiver-operating characteristic (ROC) curve analysis and comparisons were used to evaluate the ability of single diffusion parameters and combined diffusion models to distinguish between the two groups.

RESULTS

A total of six diffusion parameters (including the cortical anomalous exponent term [α_Cortex], medullary fractional anisotropy [FA_Medulla], cortical FA [FA_Cortex], cortical axial diffusivity [Da_Cortex], medullary mean diffusivity [MD_Medulla] and medullary radial diffusivity [Dr_Medulla]) were included, and 10 regression models were studied. The area under the curve (AUC) of Dr_Medulla was 0.855, comparable to that of FA_Cortex and FA_Medulla and significantly higher than that of α_Cortex, Da_Cortex and MD_Medulla. The AUC of the Model_all parameters was 0.967, comparable to that of Model_FA (0.946) and Model_DKI (0.966) and significantly higher than that of the other models. The sensitivity and specificity of Model_all parameters were 87.2% and 95%, respectively.

CONCLUSION

The Model_all parameters, Model_FA and Model_DKI were valid for differentiating between PA patients and HCs with similar differentiation efficacy and were superior to single diffusion parameters and other models.

Truly reproducible uniform estimation of the ADC with multi-b diffusion data- Application in prostate diffusion imaging

PURPOSE

The ADC is a well-established parameter for clinical diagnostic applications, but lacks reproducibility because it is also influenced by the choice diffusion weighting level. A framework is evaluated that is based on multi-b measurement over a wider range of diffusion-weighting levels and higher order tissue diffusion modeling with retrospective, fully reproducible ADC calculation.

METHODS

Averaging effect from curve fitting for various model functions at 20 linearly spaced b-values was determined by means of simulations and theoretical calculations. Simulation and patient multi-b image data were used to compare the new approach for diffusion-weighted image and ADC map reconstruction with and without Rician bias correction to an active clinical trial protocol probing three non-zero b-values.

RESULTS

Averaging effect at a certain b-value varies for model function and maximum b-value used. Images and ADC maps from the novel procedure are on-par with the clinical protocol. Higher order modeling and Rician bias correction is feasible, but comes at the cost of longer computation times.

CONCLUSIONS

Application of the new framework makes higher order modeling more feasible in a clinical setting while still providing patient images and reproducible ADC maps of adequate quality.

Sub-differentiation of PI-RADS 3 lesions in TZ by advanced diffusion-weighted imaging to aid the biopsy decision process

Background

Performing biopsy for intermediate lesions with PI-RADS 3 has always been controversial. Moreover, it is difficult to differentiate prostate cancer (PCa) and benign prostatic hyperplasia (BPH) nodules in PI-RADS 3 lesions by conventional scans, especially for transition zone (TZ) lesions. The purpose of this study is sub-differentiation of transition zone (TZ) PI-RADS 3 lesions using intravoxel incoherent motion (IVIM), stretched exponential model, and diffusion kurtosis imaging (DKI) to aid the biopsy decision process.

Methods

A total of 198 TZ PI-RADS 3 lesions were included. 149 lesions were BPH, while 49 lesions were PCa, including 37 non-clinical significant PCa (non-csPCa) lesions and 12 clinical significant PCa (csPCa) lesions. Binary logistic regression analysis was used to examine which parameters could predict PCa in TZ PI-RADS 3 lesions. The ROC curve was used to test diagnostic efficiency in distinguishing PCa from TZ PI-RADS 3 lesions, while one-way ANOVA analysis was used to examine which parameters were statistically significant among BPH, non-csPCa and csPCa.

Results

The logistic model was statistically significant (χ2 = 181.410, p<0.001) and could correctly classify 89.39% of the subjects. Parameters of fractional anisotropy (FA) (p=0.004), mean diffusion (MD) (p=0.005), mean kurtosis (MK) (p=0.015), diffusion coefficient (D) (p=0.001), and distribute diffusion coefficient (DDC) (p=0.038) were statistically significant in the model. ROC analysis showed that AUC was 0.9197 (CI 95%: 0.8736-0.9659). Sensitivity, specificity, positive predictive value and negative predictive value were 92.1%, 80.4%, 93.9% and 75.5%, respectively. FA and MK of csPCa were higher than those of non-csPCa (all p<0.05), while MD, ADC, D, and DDC of csPCa were lower than those of non-csPCa (all p<0.05).

Conclusion

FA, MD, MK, D, and DDC can predict PCa in TZ PI-RADS 3 lesions and inform the decision-making process of whether or not to perform a biopsy. Moreover, FA, MD, MK, D, DDC, and ADC may have ability to identify csPCa and non-csPCa in TZ PI-RADS 3 lesions.

Monoexponential, biexponential, stretched-exponential and kurtosis models of diffusion-weighted imaging in kidney assessment: comparison between patients with primary aldosteronism and healthy controls

PURPOSE

This study used various diffusion-weighted imaging (DWI) models (including monoexponential, biexponential, stretched-exponential and kurtosis models) in renal magnetic resonance imaging (MRI) to compare whether there were differences in each diffusion parameter between patients with primary aldosteronism (PA) and healthy volunteers.

MATERIALS AND METHODS

Twenty-two (female:male, 14:8; age, 48 ± 10 years) patients with PA and 22 age- and sex-matched healthy controls (HCs) underwent MRI examinations of the kidneys. The independent-sample t test or the Mann‒Whitney U test was used to detect differences in the diffusion metrics of the kidneys between the two groups. Univariable and multivariable linear regression were applied to analyze the correlations between diffusion parameters and the clinical indicators.

RESULTS

The mean diffusivity (MD, p < 0.001) and radial diffusivity (Dr, p < 0.001) values in the medulla were lower in the PA group than in the HC group. The medullary fractional anisotropy (FA, p < 0.001) was higher than that of HCs. The FA (p < 0.001) and axial diffusivity (Da, p < 0.001) values in the cortex were lower in the PA group. The cortical α (anomalous exponent term, p = 0.016) was higher in the PA patients than in the HCs. Linear regression analysis showed that log(plasma aldosterone concentration) and the estimated glomerular filtration rate (eGFR) were correlated with medullary FA.

CONCLUSION

The stretched-exponential model (cortical α) and the kurtosis model (FA, MD and Dr in the medulla and FA and Da in the cortex) showed significant differences between PA patients and healthy volunteers and may have potential for noninvasive renal assessment in PA patients.

Quantitative differentiation of malignant and benign thyroid nodules with multi-parameter diffusion-weighted imaging

BACKGROUND

The value of conventional magnetic resonance imaging in the differential diagnosis of thyroid nodules is limited; however, the value of multi-parameter diffusion-weighted imaging (DWI) in the quantitative evaluation of thyroid nodules has not been well determined.

AIM

To determine the utility of multi-parametric DWI including mono-exponential, bi-exponential, stretched exponential, and kurtosis models for the differentiation of thyroid lesions.

METHODS

Seventy-nine patients (62 with benign and 17 with malignant nodules) underwent multi-b value diffusion-weighted imaging of the thyroid. Multiple DWI parameters were obtained for statistical analysis.

RESULTS

Good agreement was found for diffusion parameters of thyroid nodules. Malignant lesions displayed lower diffusion parameters including apparent diffusion coefficient (ADC), the true diffusion coefficient (D), the perfusion fraction (f), the distributed diffusion coefficient (DDC), the intravoxel water diffusion heterogeneity (α) and kurtosis model-derived ADC (Dapp), and higher apparent diffusional kurtosis (Kapp) than benign entities (all P < 0.01), except for the pseudodiffusion coefficient (D*) (P > 0.05). The area under the ROC curve (AUC) of the ADC_{(0 and 1000)} was not significantly different from that of the ADC_{(0 and 2000)}, ADC_{(0 to 2000)}, ADC_{(0 to 1000)}, D, DDC, Dapp and Kapp (all P > 0.05), but was significantly higher than the AUC of D*, f and α (all P < 0.05) for differentiating benign from malignant lesions.

CONCLUSION

Multiple DWI parameters including ADC, D, f, DDC, α, Dapp and Kapp could discriminate benign and malignant thyroid nodules. The metrics including D, DDC, Dapp and Kapp provide additional information with similar diagnostic performance of ADC, combination of these metrics may contribute to differentiate benign and malignant thyroid nodules. The ADC calculated with higher b values may not lead to improved diagnostic performance.

[Feasibility of Cerebrovascular Reserve Assessment Using Stretched Exponential Model in Major Cerebral Artery Steno-occlusive Disease: Comparison with SPECT]

PURPOSE

To clarify whether diffusion-weighted imaging using stretched exponential model can assess cerebrovascular reserve (CVR) in patients with major cerebral artery steno-occlusive disease, we compared stretched exponential parameters and single-photon emission computed tomography (SPECT).

METHODS

Twenty-nine patients with unilateral major cerebral artery steno-occlusive disease (25 men and 4 women; age, 69±11 years) were analyzed in this study. The patients were divided into three groups: normal CVR (CVR≥30%), moderate CVR (10%≤CVR<30%), and severe CVR (CVR<10%). The distributed diffusion coefficient (DDC) and heterogeneity index (α) from the stretched exponential model, apparent diffusion coefficient (ADC) from the monoexponential model, and CVR and resting cerebral blood flow (CBF) from SPECT were measured in the bilateral middle cerebral artery territories, and ipsilateral-to-contralateral ratios (rDDC, rα, rADC, and rCBF) were obtained.

RESULTS

The rDDC values in severe CVR were significantly higher than those in normal CVR (P=0.003). The rDDC values were significantly negatively correlated with ipsilateral CVR (rho=-0.31, P=0.009). The rDDC values were not significantly correlated with rCBF (P=0.34).

CONCLUSION

We have shown that elevated rDDC values are associated with impaired CVR. Our results suggest that diffusion-weighted imaging using stretched exponential model has a potential to evaluate hemodynamic impairment.

Differentiation of Prostate Cancer and Stromal Hyperplasia in the Transition Zone With Monoexponential, Stretched-Exponential Diffusion-Weighted Imaging and Diffusion Kurtosis Imaging in a Reduced Number of b Values: Correlation With Whole-Mount Pathology

OBJECTIVES

The aims of the study were to explore the feasibility of generating a monoexponential model (MEM), stretched-exponential model (SEM) based diffusion-weighted imaging (DWI), and diffusion kurtosis imaging (DKI) by applying the same set of reduced b values and to compare their effectiveness in distinguishing prostate cancer from stromal hyperplasia (SH) in the transition zone (TZ) area.

METHODS

An analysis of 75 patients who underwent preoperative DWI (b values of 0, 700, 1400, 2000 s/mm2) was performed. All lesions were localized on magnetic resonance images according to whole-mount histopathological correlations. The apparent diffusion coefficient (ADC), water molecular diffusion heterogeneity index (α), distributed diffusion coefficient (DDC), mean diffusivity (MD), and mean kurtosis (MK) values were calculated and compared between the TZ cancer and SH groups. Receiver operating characteristic analysis and areas under the receiver operating characteristic curve (AUCs) were carried out for all parameters.

RESULTS

Compared with the SH group, the ADC, DDC, α, and MD values of the TZ cancer group were significantly reduced, while the MK value was significantly increased (all P < 0.05). The AUCs of the ADC, DDC, α, MD, and MK were 0.828, 0.801, 0.813, 0.822, and 0.882, respectively. The AUC of MK was significantly higher than that of the other parameters (all P < 0.05).

CONCLUSIONS

When using the reduced b-value set, all parameters from MEM, SEM, based DWI, and DKI can effectively distinguish TZ cancer from SH. Among them, DKI demonstrated potential clinical superiority over the others in TZ cancer diagnosis.

Diffusion and quantification of diffusion of prostate cancer

For assessing a cancer treatment, and for detecting and characterizing cancer, Diffusion-weighted imaging (DWI) is commonly used. The key in DWI's use extracranially has been due to the emergence of of high-gradient amplitude and multichannel coils, parallelimaging, and echo-planar imaging. The benefit has been fewer motion artefacts and high-quality prostate images.Recently, new techniques have been developed to improve the signal-to-noise ratio of DWI with fewer artefacts, allowing an increase in spatial resolution. For apparent diffusion coefficient quantification, non-Gaussian diffusion models have been proposed as additional tools for prostate cancer detection and evaluation of its aggressiveness. More recently, radiomics and machine learning for prostate magnetic resonance imaging have emerged as novel techniques for the non-invasive characterisation of prostate cancer. This review presents recent developments in prostate DWI and discusses its potential use in clinical practice.

Evaluation of microvascular invasion of hepatocellular carcinoma using whole-lesion histogram analysis with the stretched-exponential diffusion model

OBJECTIVES

To evaluate the potential role of histogram analysis of stretched exponential model (SEM) through whole-tumor volume for preoperative prediction of microvascular invasion (MVI) in single hepatocellular carcinoma (HCC).

METHODS

This study included 43 patients with pathologically proven HCCs by surgery who underwent multiple b-values diffusion-weighted imaging (DWI) and contrast-enhanced MRI.The histogram metrics of distributed diffusion coefficient (DDC) and heterogeneity index (α) from SEM were compared between HCCs with and without MVI, by using the independent t-test. Morphologic features of conventional MRI and clinical data were evaluated with chi-squared or Fisher's exact tests. Receiver operating characteristic (ROC) and multivariable logistic regression analyses were performed to evaluate the diagnostic performance of different parameters for predicting MVI.

RESULTS

The tumor size and non-smooth tumor margin were significantly associated with MVI (all p < 0.05). The mean, fifth, 25th, 50th percentiles of DDC, and the fifth percentile of ADC between HCCs with and without MVI were statistically significant differences (all p < 0.05). The histogram parameters of α showed no statistically significant differences (all p > 0.05). At multivariate analysis,the fifth percentile of DDC was independent risk factor for MVI of HCC(p = 0.006).

CONCLUSIONS

Histogram parameters DDC and ADC, but not the α value, are useful predictors of MVI. The fifth percentile of DDC was the most useful value to predict MVI of HCC.

ADVANCES IN KNOWLEDGE

There is limited literature addressing the role of SEM for evaluating MVI of HCC. Our findings suggest that histogram analysis of SEM based on whole-tumor volume can be useful for MVI prediction.

Noninvasive Differentiation of Obstructive Azoospermia and Nonobstructive Azoospermia Using Multimodel Diffusion Weighted Imaging

RATIONALE AND OBJECTIVES

To evaluate the diagnostic performance of parameters derived from multimodel diffusion weighted imaging (monoexponential, stretched-exponential diffusion weighted imaging and diffusion kurtosis imaging [DKI]) from noninvasive magnetic resonance imaging in distinguishing obstructive azoospermia (OA) from nonobstructive azoospermia (NOA).

MATERIALS AND METHODS

Forty-six patients with azoospermia were prospectively enrolled and classified into two groups (21 OA patients and 25 NOA patients). The multimodel parameters of diffusion-weighted imaging (DWI; apparent diffusion coefficient [ADC], distributed diffusion coefficient [DDC], diffusion heterogeneity [α], diffusion kurtosis diffusivity [D_app], and diffusion kurtosis coefficient [K_app]) were derived. The diagnostic performance of these parameters for the differentiation of OA and NOA patients were evaluated using receiver operating characteristic analysis. The area under the curve (AUC) was calculated to evaluate the diagnostic accuracy of each parameter.

RESULTS

All the parameters (ADC, α, DDC, D_app, and K_app) values were significantly different between OA and NOA (P < 0.001 for all). For the differentiation of OA from NOA, K_app showed the highest AUC value (0.965), followed by DDC (0.946), D_app (0.933), ADC (0.922), and α (0.887). K_app had a significantly higher AUC than the conventional ADC (P < 0.05).

CONCLUSION

Parameters derived from multimodels of DWI have the potential for the noninvasive differentiation of OA and NOA. The K_app value derived from the DKI model might serve as a useful imaging marker for the differentiation of azoospermia.

Prediction of chemotherapy response in breast cancer patients at pre-treatment using second derivative texture of CT images and machine learning

•

Textural and second derivative textural features of CT images can be used in conjunction with machine learning models to predict breast cancer response to chemotherapy prior to the start of treatment.

•

The proposed predictive model separates the patients at pre-treatment into two cohorts (responders/non-responders) with significantly different survival.

•

The proposed methodology is a step forward towards the precision oncology paradigm for breast cancer patients.

Although neoadjuvant chemotherapy (NAC) is a crucial component of treatment for locally advanced breast cancer (LABC), only about 70% of patients respond to it. Effective adjustment of NAC for individual patients can significantly improve survival rates of those resistant to standard regimens. Thus, the early prediction of NAC outcome is of great importance in facilitating a personalized paradigm for breast cancer therapeutics. In this study, quantitative computed tomography (qCT) parametric imaging in conjunction with machine learning techniques were investigated to predict LABC tumor response to NAC. Textural and second derivative textural (SDT) features of CT images of 72 patients diagnosed with LABC were analysed before the initiation of NAC to quantify intra-tumor heterogeneity. These quantitative features were processed through a correlation-based feature reduction followed by a sequential feature selection with a bootstrap 0.632+ area under the receiver operating characteristic (ROC) curve (AUC0.632+) criterion. The best feature subset consisted of a combination of one textural and three SDT features. Using these features, an AdaBoost decision tree could predict the patient response with a cross-validated AUC0.632+ accuracy, sensitivity and specificity of 0.88, 85%, 88% and 75%, respectively. This study demonstrates, for the first time, that a combination of textural and SDT features of CT images can be used to predict breast cancer response NAC prior to the start of treatment which can potentially facilitate early therapy adjustments.

Comparison of mono-exponential, bi-exponential, and stretched exponential diffusion-weighted MR imaging models in differentiating hepatic hemangiomas from liver metastases

PURPOSE

This study aims to compare the diagnostic values of mono-exponential, bi-exponential, and stretched exponential diffusion-weighted imaging (DWI) in differentiating hepatic hemangiomas and liver metastases.

METHOD

This prospective study was approved by our institutional review board, and written informed consent was obtained from all patients. In this study, 244 patients with known or suspected liver disease underwent magnetic resonance imaging. Among them, 37 patients who had focal hepatic lesions with a maximum diameter of ≥10 mm were evaluated. Using home-built software, two radiologists measured the DWI parameters of hepatic lesions for the three models: the apparent diffusion coefficient (ADC) from a mono-exponential model; the true diffusion coefficient (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f) from a bi-exponential model; and the distributed diffusion coefficient (DDC) and water molecular diffusion heterogeneity index (α) from a stretched exponential model. The parameters were compared between hepatic hemangiomas and liver metastases.

RESULTS

In total, 64 focal hepatic lesions were evaluated, of which 22 were identified to be hepatic hemangiomas and 42 were liver metastases. ADC, D, f, and DDC values were significantly lower in liver metastases than in hepatic hemangiomas (P <  0.0001, <  0.0001, 0.015, and <  0.0001, respectively); whereas, the α value was significantly higher in liver metastases than in hepatic hemangiomas (P =  0.028). The areas under the ROC curve (AUCs) for differentiating hepatic hemangiomas and liver metastases in ADC, D, D*, f, DDC, and α were 0.940, 0.908, 0.608, 0.686, 0.952, and 0.667, respectively. The AUC values of ADC and DDC were significantly greater than those of D* (P <  0.0001), f (P =  0.0001), and α values (P =  0.0001).

CONCLUSION

ADC and DDC values from the mono-exponential and stretched exponential models could be considered as quantitative imaging biomarkers for differentiating hepatic hemangiomas and liver metastases.

Noninvasive DW-MRI metrics for staging hepatic fibrosis and grading inflammatory activity in patients with chronic hepatitis B

PURPOSE

To assess the value of various diffusion parameters obtained from monoexponential, biexponential, and stretched-exponential diffusion-weighted imaging (DWI) models for staging hepatic fibrosis (HF) and grading inflammatory activity in patients with chronic hepatitis B (CHB).

METHODS

82 patients with CHB and 30 healthy volunteers underwent DWI with 13 b-values on a 3T MRI unit. The standard apparent diffusion coefficient (ADC_st) was calculated using a monoexponential model. The true diffusion coefficient (D_t), pseudo-diffusion coefficient (D_p), and perfusion fraction (f) were calculated using a biexponential model. The distributed diffusion coefficient (DDC) and water-molecule diffusion heterogeneity index (α) were calculated using a stretched-exponential model. Receiver operating characteristic (ROC) curves were performed for diffusion parameters to compare the diagnosis performance.

RESULTS

The distributions of hepatic fibrosis stages and the inflammatory activity grades (METAVIR scoring system) were as follows: F0, n = 1; F1, n = 16; F2, n = 31; F3, n = 19; and F4, n = 15. A0, n = 1; A1, n = 14; A2, n = 46; and A3, n = 21. ADC_st, D_t and DDC values showed negative correlation with the fibrosis stage (r = - 0.418, - 0.717 and - 0.630, all P < 0.001) and the inflammatory activity grade (r = - 0.514, - 0.626 and - 0.550, all P < 0.001). The area under the ROC curve (AUC) of D_t (AUC = 0.854, 0.881) and DDC (AUC = 0.794, 0.834) were significantly higher than that of ADC_st (AUC = 0.637, 0.717) in discriminating significant fibrosis (≥ F2) and advanced fibrosis (≥ F3) (all P < 0.05). Although D_t (AUC = 0.867, 0.836) and DDC (AUC = 0.810, 0.808) showed higher AUCs than ADC_st (AUC = 0.767, 0.803), there was no significant difference in their ability in detecting inflammatory activity grade ≥ A2/A3 (P > 0.05).

CONCLUSIONS

D_t and DDC are promising indicators and outperform ADC_st for staging HF. While both D_t and DDC have similar diagnostic performance compared with ADC_st for grading inflammatory activity.

Predicting the effects of radiotherapy based on diffusion kurtosis imaging in a xenograft mouse model of esophageal carcinoma

The aim of the present study was to assess the predictive value of diffusion kurtosis imaging (DKI) on the effects of radiotherapy in a xenograft model of esophageal cancer. A total of 40 tumor-bearing mice, established by injection of Eca-109 cells in nude mice, were used. The experimental group (n=24) received a single dose of 15 Gy (6 MV by X-ray), and the control group (n=16) did not receive any treatment. Tumor volume, apparent diffusion coefficient (ADC), mean kurtosis (MK) and mean diffusivity (MD) of the two groups were compared, and the expression of aquaporin (AQP) 3 and necrosis ratio at matched time points in xenografts were also observed. There was a significant difference between the two groups from the 7th day of radiotherapy onwards; the xenograft volume of the experimental group was significantly smaller compared with the control group (P<0.05). On the 3rd day, the ADC and MD of the experimental group was significantly higher compared with the control group, and MK was significantly lower compared with the control group (P<0.05). On the 3rd day, AQP3 expression in the experimental group was lower compared with the control group, and the proportion of necrotic cells was higher compared with the control group (P<0.05). Single large fraction dose radiotherapy inhibited the growth of a xenografted esophageal tumor. Changes in ADC, MK and MD were observed prior to morphological changes in the tumor. The change in AQP3 expression and necrosis ratio was in also agreement with the DKI parameters assessed. DKI may thus provide early predictive ability on the effect of radiotherapy in esophageal carcinoma.

Comparing mono-exponential, bi-exponential, and stretched-exponential diffusion-weighted MR imaging for stratifying non-alcoholic fatty liver disease in a rabbit model

OBJECTIVES

To compare diffusion parameters obtained from mono-exponential, bi-exponential, and stretched-exponential diffusion-weighted imaging (DWI) in stratifying non-alcoholic fatty liver disease (NAFLD).

METHODS

Thirty-two New Zealand rabbits were fed a high-fat/cholesterol or standard diet to obtain different stages of NAFLD before 12 b-values (0-800 s/mm2) DWI. The apparent diffusion coefficient (ADC) from the mono-exponential model; pure water diffusion (D), pseudo-diffusion (D*), and perfusion fraction (f) from bi-exponential DWI; and distributed diffusion coefficient (DDC) and water molecular diffusion heterogeneity index (α) from stretched-exponential DWI were calculated for hepatic parenchyma. The goodness of fit of the three models was compared. NAFLD severity was pathologically graded as normal, simple steatosis, borderline, and non-alcoholic steatohepatitis (NASH). Spearman rank correlation analysis and receiver operating characteristic curves were used to assess NAFLD severity.

RESULTS

Upon comparison, the goodness of fit chi-square from stretched-exponential fitting (0.077 ± 0.012) was significantly lower than that for the bi-exponential (0.110 ± 0.090) and mono-exponential (0.181 ± 0.131) models (p < 0.05). Seven normal, 8 simple steatosis, 6 borderline, and 11 NASH livers were pathologically confirmed from 32 rabbits. Both α and D increased with increasing NAFLD severity (r = 0.811 and 0.373, respectively; p < 0.05). ADC, f, and DDC decreased as NAFLD severity increased (r = - 0.529, - 0.717, and - 0.541, respectively; p < 0.05). Both α (area under the curve [AUC] = 0.952) and f (AUC = 0.931) had significantly greater AUCs than ADC (AUC = 0.727) in the differentiation of NASH from borderline or less severe groups (p < 0.05).

CONCLUSIONS

Stretched-exponential DWI with higher fitting efficiency performed, as well as bi-exponential DWI, better than mono-exponential DWI in the stratification of NAFLD severity.

KEY POINTS

• Stretched-exponential diffusion model fitting was more reliable than the bi-exponential and mono-exponential diffusion models (p = 0.039 and p < 0.001, respectively). • As NAFLD severity increased, the diffusion heterogeneity index (α) increased, while the perfusion fraction (f) decreased (r = 0.811, - 0.717, p < 0.05). • Both α and f showed superior NASH diagnostic performance (AUC = 0.952, 0.931) compared with ADC (AUC = 0.727, p < 0.05).

Potentialities of multi-b-values diffusion-weighted imaging for predicting efficacy of concurrent chemoradiotherapy in cervical cancer patients

BACKGROUND

To testify whether multi-b-values diffusion-weighted imaging (DWI) can be used to ultra-early predict treatment response of concurrent chemoradiotherapy (CCRT) in cervical cancer patients and to assess the predictive ability of concerning parameters.

METHODS

Fifty-three patients with biopsy proved cervical cancer were retrospectively recruited in this study. All patients underwent pelvic multi-b-values DWI before and at the 3rd day during treatment. The apparent diffusion coefficient (ADC), true diffusion coefficient (D_slow), perfusion-related pseudo-diffusion coefficient (D_fast), perfusion fraction (f), distributed diffusion coefficient (DDC) and intravoxel diffusion heterogeneity index(α) were generated by mono-exponential, bi-exponential and stretched exponential models. Treatment response was assessed based on Response Evaluation Criteria in Solid Tumors (RECIST v1.1) at 1 month after the completion of whole CCRT. Parameters were compared using independent t test or Mann-Whitney U test as appropriate. Receiver operating characteristic (ROC) curves was used for statistical evaluations.

RESULTS

ADC-T0 (p = 0.02), D_slow-T0 (p <  0.01), DDC-T0 (p = 0.03), ADC-T1 (p <  0.01), D_slow-T1 (p <  0.01), ΔADC (p = 0.04) and Δα (p <  0.01) were significant lower in non-CR group patients. ROC analyses showed that ADC-T1 and Δα exhibited high prediction value, with area under the curves of 0.880 and 0.869, respectively.

CONCLUSIONS

Multi-b-values DWI can be used as a noninvasive technique to assess and predict treatment response in cervical cancer patients at the 3rd day of CCRT. ADC-T1 and Δα can be used to differentiate good responders from poor responders.

A priori prediction of tumour response to neoadjuvant chemotherapy in breast cancer patients using quantitative CT and machine learning

Response to Neoadjuvant chemotherapy (NAC) has demonstrated a high correlation to survival in locally advanced breast cancer (LABC) patients. An early prediction of responsiveness to NAC could facilitate treatment adjustments on an individual patient basis that would be expected to improve treatment outcomes and patient survival. This study investigated, for the first time, the efficacy of quantitative computed tomography (qCT) parametric imaging to characterize intra-tumour heterogeneity and its application in predicting tumour response to NAC in LABC patients. Textural analyses were performed on CT images acquired from 72 patients before the start of chemotherapy to determine quantitative features of intra-tumour heterogeneity. The best feature subset for response prediction was selected through a sequential feature selection with bootstrap 0.632 + area under the receiver operating characteristic (ROC) curve (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{A}\mathrm{U}\mathrm{C}}_{0.632+}$$\end{document}AUC0.632+) as a performance criterion. Several classifiers were evaluated for response prediction using the selected feature subset. Amongst the applied classifiers an Adaboost decision tree provided the best results with cross-validated \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{A}\mathrm{U}\mathrm{C}}_{0.632+}$$\end{document}AUC0.632+, accuracy, sensitivity and specificity of 0.89, 84%, 80% and 88%, respectively. The promising results obtained in this study demonstrate the potential of the proposed biomarkers to be used as predictors of LABC tumour response to NAC prior to the start of treatment.

Multiple mathematical models of diffusion-weighted imaging for endometrial cancer characterization: Correlation with prognosis-related risk factors

PURPOSE

To investigate mono-exponential, bi-exponential, and stretched-exponential models of diffusion-weighted imaging (DWI) for evaluation of prognosis-related risk factors of endometrial cancer (EC).

METHOD

Sixty-one consecutive patients with EC who preoperatively underwent pelvic MRI with multiple b value DWI between September 2016 and May 2018 were enrolled. The apparent-diffusion-coefficient (ADC), bi-exponential model parameters (D, D* and f) and stretched-exponential model parameters (DDC and α) were measured and compared to analyze the following prognosis-related risk factors confirmed by pathology: histological grade, depth of myometrial invasion, cervical stromal infiltration (CSI) and lymphovascular invasion (LVSI). A stepwise multilvariate logistic regression and the receiver operating characteristic (ROC) curves were performed for further statistical analysis.

RESULTS

Lower ADC, D, f, and DDC were observed in tumor with high grade compared with a low-grade group, and the largest area under curve (AUC) was obtained when combining f and DDC values. ADC, D, f, DDC, and α were significantly different in patients with deep myometrial invasion (DMI) compared to those without DMI; the combination of f, DDC and α showed the highest AUC. Significantly different ADC and f were found between patients' presence and absence CSI; the f values showed the highest diagnostic performance with an AUC of 0.825. Regarding the LVSI, ADC, D*, f, and DDC were significantly lower in tumors with LVSI compared to those without LVSI; the combination of f and DDC showed the largest AUC.

CONCLUSION

Multiple mathematical DWI models are a useful approach for the prediction of prognosis-related risk factors in EC.

Multiple b-value diffusion-weighted imaging in differentiating benign from malignant breast lesions: comparison of conventional mono-, bi- and stretched exponential models

AIM

To prospectively evaluate multiple b-value diffusion-weighted imaging (DWI) in differentiating malignant from benign breast lesions.

MATERIALS AND METHODS

The study cohort included 103 patients who underwent 3 T magnetic resonance imaging (MRI). The conventional sequences included T1-weighted dynamic contrast-enhanced, T1-weighted and T2-weighted fat-suppressed sequences, single b-value (b=0, 1000 s/mm²) DWI, and multiple b-values (12 values, from 0 to 3,000 s/mm²) DWI. Pathological diagnosis of breast lesions was based on the latest World Health Organization (WHO) guide on the pathology and immunohistochemistry of the breast. SPSS Statistics V19.0 was used for the statistics analysis.

RESULTS

The following parameters were calculated: apparent diffusion coefficient (ADC), tissue diffusivity (D), perfusion fraction (f), pseudo-diffusion coefficient (D∗), distributed diffusion coefficient (DDC), and alpha (α) by the same radiologist twice (interval time of 3 months). There was good inter/intra-observer agreement for each of the parameters. The D, D∗, f, DDC, and α values were significantly different among malignant tumours, benign lesions, and normal breast tissue (p<0.05).

CONCLUSION

D, f, DDC, α, and ADC values have good sensitivity and specificity, respectively. In addition, the combined use of D and f or DDC and α has good diagnostic performance. Thus, the applications of the new multi-b DWI variables or combined variables are promising.

Comparison of the Diagnostic Value of Mono-exponential, Bi-exponential, and Stretched Exponential Signal Models in Diffusion-weighted MR Imaging for Differentiating Benign and Malignant Hepatic Lesions

PURPOSE

To compare the diagnostic value of mono-exponential, bi-exponential, and stretched exponential diffusion-weighted imaging (DWI) for differentiating benign and malignant hepatic lesions.

METHODS

This prospective study was approved by our Institutional Review Board and the patients provided written informed consent. Magnetic resonance imaging was acquired for 56 patients with suspected liver disease. This identified 90 focal liver lesions with a maximum diameter >10 mm, of which 47 were benign and 43 were malignant. Using home-built software, two radiologists measured the DWI parameters of hepatic lesions for three models: the apparent diffusion coefficient (ADC) from a mono-exponential model; the true diffusion coefficient (D), pseudo-diffusion coefficient (D^*), and perfusion fraction (f) from a bi-exponential model; and the distributed diffusion coefficient (DDC) and water molecular diffusion heterogeneity index (α) from a stretched exponential model. The parameters were compared between benign and malignant hepatic lesions.

RESULTS

ADC, D, D^*, f, and DDC values were significantly lower for malignant hepatic lesions than for benign lesions (P < 0.0001-0.03). Although logistic regression analysis demonstrated that DDC was the only statistically significant parameter for differentiating benign and malignant lesions (P = 0.039), however, the areas under the receiver operating characteristic curve for differentiating benign and malignant lesions were comparable between ADC (0.98) and DDC (0.98) values.

CONCLUSION

DDC values obtained from the stretched exponential model could be also used as a quantitative imaging biomarker for differentiating benign and malignant hepatic lesions, however, the diagnostic performance was comparable with ADC values.

The value of DWI in predicting the response to synchronous radiochemotherapy for advanced cervical carcinoma: comparison among three mathematical models

Background

Diffusion weighted imaging(DWI) mode mainly includes intravoxel incoherent motion (IVIM), stretched exponential model (SEM) and Gaussian diffusion model, but it is still unclear which mode is the most valuable in predicting the response to radiochemotherapy for cervical cancer. This study aims to compare the values of three mathematical models in predicting the response to synchronous radiochemotherapy for cervical cancer.

Methods

Eighty-four patients with cervical cancer were enrolled into this study. They underwent DWI examination by using 12 b-values prior to treatment. The imaging parameters were calculated on the basis of IVIM, SEM and Gaussian diffusion models respectively. The imaging parameters derived from three mathematical modes were compared between responders and non-responders groups. The repeatability of each imaging parameter was assessed.

Results

The ADC, D or DDC value was lower in responders than in non-responders groups (P = 0.03, 0.02, 0.01). The α value was higher in responders group than in non-responders group (P = 0.03). DDC had the largest area under curves (AUC) (=0.948) in predicting the response to treatment. The imaging parameters derived from SEM had better repeatability (CCC for DDC and α were 0.969 and 0.924 respectively) than that derived from other exponential models.

Conclusion

Three exponential modes of DWI are useful for predicting the response to radiochemotherapy for cervical cancer, and SEM may be used as a potential optimal model for predicting treatment effect.

Histogram analysis from stretched exponential model on diffusion-weighted imaging: evaluation of clinically significant prostate cancer

OBJECTIVE

To evaluate the usefulness of histogram analysis of stretched exponential model (SEM) on diffusion-weighted imaging in evaluating clinically significant prostate cancer (CSC).

METHODS

A total of 85 patients with prostate cancer underwent 3 T multiparametric MRI, followed by radical prostatectomy. Histogram parameters of the tumor from the SEM [distributed diffusion coefficient (DDC) and α] and the monoexponential model [MEM; apparent diffusion coefficient (ADC)] were evaluated. The associations between parameters and Gleason score or Prostate Imaging Reporting and Data System v. 2 were evaluated. The area under the receiver operating characteristics curve was calculated to evaluate diagnostic performance of parameters in predicting CSC.

RESULTS

The values of histogram parameters of DDC and ADC were significantly lower in patients with CSC than in patients without CSC (p < 0.05), except for skewness and kurtosis. The value of the 25th percentile of α was significantly lower in patients with CSC than in patients without CSC (p = 0.014). Histogram parameters of ADC and DDC had significant weak to moderate negative associations with Gleason score or Prostate Imaging Reporting and Data System v. 2 (p < 0.001), except for skewness and kurtosis. For predicting CSC, the area under the curves of mean ADC (0.856), 50th percentile DDC (0.852), and 25th percentile α (0.707) yielded the highest values compared to other histogram parameters from each group.

CONCLUSION

Histogram analysis of the SEM on diffusion-weighted imaging may be a useful quantitative tool for evaluating CSC. However, the SEM did not outperform the MEM.

ADVANCES IN KNOWLEDGE

Histogram parameters of SEM may be useful for evaluating CSC.

Non-Gaussian models of diffusion weighted imaging for detection and characterization of prostate cancer: a systematic review and meta-analysis

The importance of Diffusion Weighted Imaging (DWI) in prostate cancer (PCa) diagnosis have been widely handled in literature. In the last decade, due to the mono-exponential model limitations, several studies investigated non-Gaussian DWI models and their utility in PCa diagnosis. Since their results were often inconsistent and conflicting, we performed a systematic review of studies from 2012 examining the most commonly used Non-Gaussian DWI models for PCa detection and characterization. A meta-analysis was conducted to assess the ability of each Non-Gaussian model to detect PCa lesions and distinguish between low and intermediate/high grade lesions. Weighted mean differences and 95% confidence intervals were calculated and the heterogeneity was estimated using the I2 statistic. 29 studies were selected for the systematic review, whose results showed inconsistence and an unclear idea about the actual usefulness and the added value of the Non-Gaussian model parameters. 12 studies were considered in the meta-analyses, which showed statistical significance for several non-Gaussian parameters for PCa detection, and to a lesser extent for PCa characterization. Our findings showed that Non-Gaussian model parameters may potentially play a role in the detection and characterization of PCa but further studies are required to identify a standardized DWI acquisition protocol for PCa diagnosis.

Evaluation of intratumoral heterogeneity by using diffusion kurtosis imaging and stretched exponential diffusion-weighted imaging in an orthotopic hepatocellular carcinoma xenograft model

Background

To investigate the value of diffusion kurtosis imaging (DKI) and diffusion-weighted imaging (DWI) with a stretched exponential model (SEM) in the evaluation of tumor heterogeneity in an orthotopic hepatocellular carcinoma (HCC) xenograft model.

Methods

Thirty orthotopic HCC xenograft nude mice models were established and randomly divided into two groups, the sorafenib induction group (n=15) and control group (n=15). Every mouse in each group underwent MRI with DKI and SEM on a 1.5T MR scanner at 7, 14, and 21 days after sorafenib intervention. DKI and SEM parameters including mean kurtosis (MK), mean diffusivity (MD), α, and distributed diffusion coefficient (DDC) were measured, calculated, and compared between the two groups and among different time points. Sequential correlations between histopathological results including necrotic fraction (NF), micro-vessel density (MVD), Ki-67 index, standard deviation (SD), and kurtosis from hematoxylin-eosin staining, and DKI and SEM parameters were analyzed.

Results

MK, MD, and DDC of HCC in the sorafenib induction group were significantly higher than those in the control group at each time point (P<0.05), while α was significantly lower (P<0.05). Significantly positive correlations were found between MK and NF (r=0.693, P=0.010), SD (r =0.785, P=0.003), kurtosis (r=0.779, P=0.003), between MD and NF (r=0.794, P=0.003), SD (r=0.629, P=0.020), kurtosis (r=0.645, P=0.018), and between DDC and NF (r=0.800, P=0.003), SD (r=0.636, P=0.020), kurtosis (r=0.664, P=0.016), and significantly negative correlations were observed between α and NF (r=-0.704, P=0.009), SD (r=-0.754, P=0.003), and kurtosis (r=-0.792, P=0.003) in the sorafenib induction group.

Conclusions

DKI and SEM parameters may be potentially useful for evaluating intratumoral heterogeneity in HCC.

Monoexponential, Biexponential, and stretched-exponential models using diffusion-weighted imaging: A quantitative differentiation of breast lesions at 3.0T

BACKGROUND

Diffusion-weighted imaging (DWI) plays an important role in the differentiation of malignant and benign breast lesions.

PURPOSE

To investigate the utility of various diffusion parameters obtained from monoexponential, biexponential, and stretched-exponential DWI models in the differential diagnosis of breast lesions.

STUDY TYPE

Prospective.

POPULATION

Sixty-one patients (age range: 25-68 years old; mean age: 46 years old) with 31 malignant lesions, 42 benign lesions, and 28 normal breast tissues diagnosed initially by clinical palpation, ultrasonography, or conventional mammography were enrolled in the study from January to September 2016.

FIELD STRENGTH

3.0T MR scanner, T₁ WI, T₂ WI, DWI (conventional and multi-b values), dynamic contrast-enhanced.

ASSESSMENT

The apparent diffusion coefficient (ADC) was calculated by monoexponential analysis. The diffusion coefficient (ADC_slow ), pseudodiffusion coefficient (ADC_fast ), and perfusion fraction (f) were calculated using the biexponential model. The distributed diffusion coefficient (DDC) and water molecular diffusion heterogeneity index (α) were obtained using a stretched-exponential model. All parameters were compared for malignant tumors, benign tumors, and normal breast tissues. A receiver operating characteristic curve was used to compare the ability of these parameters, in order to differentiate benign and malignant breast lesions.

STATISTICAL TESTS

All statistical analyses were performed using statistical software (SPSS).

RESULTS

ADC, ADC_slow , f, DDC, and α values were significantly lower in malignant tumors when compared with normal breast tissues and benign tumors (P < 0.05). However, ADC and f had higher area under the receiver operating characteristic curve (AUC) values (0.889 and 0.919, respectively).

DATA CONCLUSION

The parameters derived from the biexponential and stretched-exponential DWI could provide additional information for differentiating between benign and malignant breast tumors when compared with conventional diffusion parameters.

LEVEL OF EVIDENCE

4 Technical Efficacy: Stage 4 J. Magn. Reson. Imaging 2019;50:1461-1467.

Characterization of focal liver lesions using the stretched exponential model: comparison with monoexponential and biexponential diffusion-weighted magnetic resonance imaging

OBJECTIVE

To compare the stretched exponential model of diffusion-weighted imaging (DWI) with monoexponential and biexponential models in terms of the ability to characterize focal liver lesions (FLLs).

METHODS

This retrospective study included 180 patients with FLLs who underwent magnetic resonance imaging including DWI with nine b values at 3.0 T. The distributed diffusion coefficient (DDC) and intravoxel diffusion heterogeneity index (α) from a stretched exponential model; true diffusion coefficient (D_t), pseudo-diffusion coefficient (D_p), and perfusion fraction (f) from a biexponential model; and apparent diffusion coefficient (ADC) were calculated for each lesion. Diagnostic performances of the parameters were assessed through receiver operating characteristic (ROC) analysis. For 20 patients with treated hepatic metastases, the correlation between the DWI parameters and the percentage of tumor necrosis on pathology was evaluated using the Spearman correlation coefficient.

RESULTS

DDC had the highest area under the ROC curve (AUC, 0.905) for differentiating malignant from benign lesions, followed by D_t (0.903) and ADC (0.866), without significant differences among them (DDC vs. D_t, p = 0.946; DDC vs. ADC, p = 0.157). For distinguishing hypovascular from hypervascular lesions, and hepatocellular carcinoma from metastasis, f  had a significantly higher AUC than the other DWI parameters (p < 0.05). The α had the strongest correlation with the degree of tumor necrosis (ρ = 0.655, p = 0.002).

CONCLUSION

The DDC from stretched exponential model of DWI demonstrated excellent diagnostic performance for differentiating malignant from benign FLLs. The α is promising for evaluating the degree of necrosis in treated metastases.

KEY POINTS

• The stretched exponential DWI model is valuable for characterizing focal liver lesions. • The DDC from stretched exponential model shows excellent performance for differentiating malignant from benign focal liver lesions. • The α from stretched exponential model is promising for evaluating the degree of necrosis in hepatic metastases after chemotherapy.

Voxel-wise prostate cell density prediction using multiparametric magnetic resonance imaging and machine learning

BACKGROUND

There are currently no methods to estimate cell density in the prostate. This study aimed to develop predictive models to estimate prostate cell density from multiparametric magnetic resonance imaging (mpMRI) data at a voxel level using machine learning techniques.

MATERIAL AND METHODS

In vivo mpMRI data were collected from 30 patients before radical prostatectomy. Sequences included T2-weighted imaging, diffusion-weighted imaging and dynamic contrast-enhanced imaging. Ground truth cell density maps were computed from histology and co-registered with mpMRI. Feature extraction and selection were performed on mpMRI data. Final models were fitted using three regression algorithms including multivariate adaptive regression spline (MARS), polynomial regression (PR) and generalised additive model (GAM). Model parameters were optimised using leave-one-out cross-validation on the training data and model performance was evaluated on test data using root mean square error (RMSE) measurements.

RESULTS

Predictive models to estimate voxel-wise prostate cell density were successfully trained and tested using the three algorithms. The best model (GAM) achieved a RMSE of 1.06 (± 0.06) × 10³ cells/mm² and a relative deviation of 13.3 ± 0.8%.

CONCLUSION

Prostate cell density can be quantitatively estimated non-invasively from mpMRI data using high-quality co-registered data at a voxel level. These cell density predictions could be used for tissue classification, treatment response evaluation and personalised radiotherapy.

A comparative study of Gaussian and non-Gaussian diffusion models for differential diagnosis of prostate cancer with in-bore transrectal MR-guided biopsy as a pathological reference

Background Although several studies have been reported on evaluating the performance of Gaussian and different non-Gaussian diffusion models on prostate cancer, few studies have been reported on the comparison of different models on differential diagnosis for prostate cancer. Purpose To compare the utility of various metrics derived from monoexponential model (MEM), biexponential model (BEM), stretched-exponential model (SEM) based diffusion-weighted imaging (DWI) and diffusion kurtosis imaging (DKI) in the differential diagnosis of prostate cancer. Material and Methods Thirty-three patients underwent magnetic resonance imaging (MRI) examination. Multi-b value and multi-direction DWIs were performed. In-bore MR-guided biopsy was performed. Apparent diffusion coefficient (ADC), pure molecular diffusion (ADC_slow), pseudo-diffusion coefficient (ADC_fast), perfusion fraction (f), water molecular diffusion heterogeneity index (α), distributed diffusion coefficient (DDC), non-Gaussian diffusion coefficient (MD), and mean kurtosis (MK) values were calculated and compared between cancerous and non-cancerous groups. Receiver operating characteristic (ROC) analysis was performed for all parameters and models. Results ADC, ADC_slow, DDC, and MD values were significantly lower while MK value was significantly higher in prostate cancer than those of prostatitis and benign prostatic hyperplasia. ADC, ADC_slow, DDC, MD, and MK could discriminate between tumor and non-tumorous lesions (area under the curve, 0.856, 0.835, 0.866, 0.918, and 0.937, respectively). MK was superior to ADC in the discrimination of prostate cancer. DKI was superior to MEM in the discrimination of prostate cancer. Conclusions Parameters derived from both Gaussian and non-Gaussian models could characterize prostate cancer. DKI may be advantageous than DWI for detection of prostate cancer.

Diffusion MRI of cancer: From low to high b-values

Following its success in early detection of cerebral ischemia, diffusion-weighted imaging (DWI) has been increasingly used in cancer diagnosis and treatment evaluation. These applications are propelled by the rapid development of novel diffusion models to extract biologically valuable information from diffusion-weighted MR signals, and significant advances in MR hardware that has enabled image acquisition with high b-values. This article reviews recent technical developments and clinical applications in cancer imaging using DWI, with a special emphasis on high b-value diffusion models. The article is organized in four sections. First, we provide an overview of diffusion models that are relevant to cancer imaging. The model parameters are discussed in relation to three tissue properties-cellularity, vascularity, and microstructures. An emphasis is placed on characterization of microstructural heterogeneity, given its novelty and close relevance to cancer. Second, we illustrate diffusion MR clinical applications in each of the following three categories: 1) cancer detection and diagnosis; 2) cancer grading, staging, and classification; and 3) cancer treatment response prediction and evaluation. Third, we discuss several practical issues, including selection of image acquisition parameters, reproducibility and reliability, motion management, image distortion, etc., that are commonly encountered when applying DWI to cancer in clinical settings. Lastly, we highlight a few ongoing challenges and provide some possible future directions, particularly in the area of establishing standards via well-organized multicenter clinical trials to accelerate clinical translation of advanced DWI techniques to improving cancer care on a large scale. Level of Evidence: 5 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:23-40.

A biomimetic tumor tissue phantom for validating diffusion-weighted MRI measurements

PURPOSE

To develop a biomimetic tumor tissue phantom which more closely reflects water diffusion in biological tissue than previously used phantoms, and to evaluate the stability of the phantom and its potential as a tool for validating diffusion-weighted (DW) MRI measurements.

METHODS

Coaxial-electrospraying was used to generate micron-sized hollow polymer spheres, which mimic cells. The bulk structure was immersed in water, providing a DW-MRI phantom whose apparent diffusion coefficient (ADC) and microstructural properties were evaluated over a period of 10 months. Independent characterization of the phantom's microstructure was performed using scanning electron microscopy (SEM). The repeatability of the construction process was investigated by generating a second phantom, which underwent high resolution synchrotron-CT as well as SEM and MR scans.

RESULTS

ADC values were stable (coefficients of variation (CoVs) < 5%), and varied with diffusion time, with average values of 1.44 ± 0.03 µm2 /ms (Δ = 12 ms) and 1.20 ± 0.05 µm2 /ms (Δ = 45 ms). Microstructural parameters showed greater variability (CoVs up to 13%), with evidence of bias in sphere size estimates. Similar trends were observed in the second phantom.

CONCLUSION

A novel biomimetic phantom has been developed and shown to be stable over 10 months. It is envisaged that such phantoms will be used for further investigation of microstructural models relevant to characterizing tumor tissue, and may also find application in evaluating acquisition protocols and comparing DW-MRI-derived biomarkers obtained from different scanners at different sites. Magn Reson Med 80:147-158, 2018. © 2017 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Histogram analysis of stretched-exponential and monoexponential diffusion-weighted imaging models for distinguishing low and intermediate/high gleason scores in prostate carcinoma

BACKGROUND

Noninvasive measures to evaluate the aggressiveness of prostate carcinoma (PCa) may benefit patients.

PURPOSE

To assess the value of stretched-exponential and monoexponential diffusion-weighted imaging (DWI) for predicting the aggressiveness of PCa.

STUDY TYPE

Retrospective study.

SUBJECTS

Seventy-five patients with PCa.

FIELD STRENGTH

3T DWI examinations were performed using b-values of 0, 500, 1000, and 2000 s/mm² .

ASSESSMENT

The research were based on entire-tumor histogram analysis and the reference standard was radical prostectomy.

STATISTICAL TESTS

The correlation analysis was programmed with Spearman's rank-order analysis between the histogram variables and Gleason grade group (GG). Receiver operating characteristic (ROC) regression was used to analyze the ability of these histogram variables to differentiate low-grade (LG) from intermediate/high-grade (HG) PCa.

RESULTS

The percentiles and mean of apparent diffusion coefficient (ADC) and distributed diffusion coefficient (DDC) were correlated with GG (ρ: 0.414-0.593), while there was no significant relation among α value, skewnesses, and kurtosises with GG (ρ:0.034-0.323). HG tumors (ADC:484 ± 136, 592 ± 139, 670 ± 144, 788 ± 146, 895 ± 141 mm² /s; DDC: 410 ± 142, 532 ± 172, 666 ± 193, 786 ± 196, 914 ± 181 mm² /s) had lower values in the 10^th , 25^th , 50^th , 75^th percentiles and means than LG tumors (ADC: 644 ± 779, 737 ± 84, 836 ± 83, 919 ± 82, 997 ± 107 mm² /s; DDC: 552 ± 82, 680 ± 94, 829 ± 112, 931 ± 106, 1045 ± 100 mm² /s). However, there was no difference between LG and HG tumors in α value (0.671 ± 0.041 vs. 0.633 ± 0.114), kurtosises (ADC 0.09 vs. 0.086; DDC -0.033 vs. -0.317), or skewnesses (ADC -0.036 vs. 0.073; DDC -0.063 vs. 0.136). The above statistics were P < 0.01. ADC10 with AUC = 0.840 and DDC10 with AUC = 0.799 were similar in discriminating between LG and HG PCa at P < 0.05.

DATA CONCLUSION

Histogram variables of DDC and ADC may predict the aggressiveness of PCa, while α value does not. The abilities of ADC10 and DDC10 to discriminate LG from HG tumors were similar, and both better than their respective means.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 1 J. MAGN. RESON. IMAGING 2018;48:491-498.

Advanced Diffusion-weighted Imaging Modeling for Prostate Cancer Characterization: Correlation with Quantitative Histopathologic Tumor Tissue Composition-A Hypothesis-generating Study

Purpose To correlate quantitative diffusion-weighted imaging (DWI) parameters derived from conventional monoexponential DWI, stretched exponential DWI, diffusion kurtosis imaging (DKI), and diffusion-tensor imaging (DTI) with quantitative histopathologic tumor tissue composition in prostate cancer in a preliminary hypothesis-generating study. Materials and Methods This retrospective institutional review board-approved study included 24 patients with prostate cancer (mean age, 63 years) who underwent magnetic resonance (MR) imaging, including high-b-value DWI and DTI at 3.0 T, before prostatectomy. The following parameters were calculated in index tumors and nontumoral peripheral zone (PZ): apparent diffusion coefficient (ADC) obtained with monoexponential fit (ADC_ME), ADC obtained with stretched exponential modeling (ADC_SE), anomalous exponent (α) obtained at stretched exponential DWI, ADC obtained with DKI modeling (ADC_DKI), kurtosis with DKI, ADC obtained with DTI (ADC_DTI), and fractional anisotropy (FA) at DTI. Parameters in prostate cancer and PZ were compared by using paired Student t tests. Pearson correlations between tumor DWI and quantitative histologic parameters (nuclear, cytoplasmic, cellular, stromal, luminal fractions) were determined. Results All DWI parameters were significantly different between prostate cancer and PZ (P < .012). ADC_ME, ADC_SE, and ADC_DKI all showed significant negative correlation with cytoplasmic and cellular fractions (r = -0.546 to -0.435; P < .034) and positive correlation with stromal fractions (r = 0.619-0.669; P < .001). ADC_DTI and FA showed correlation only with stromal fraction (r = 0.512 and -0.413, respectively; P < .045). α did not correlate with histologic parameters, whereas kurtosis showed significant correlations with histopathologic parameters (r = 0.487, 0.485, -0.422 for cytoplasmic, cellular, and stromal fractions, respectively; P < .040). Conclusion Advanced DWI methods showed significant correlations with histopathologic tissue composition in prostate cancer. These findings should be validated in a larger study. ^© RSNA, 2017 Online supplemental material is available for this article. An earlier incorrect version of this article appeared online. This article was corrected on November 10, 2017.

Design and validation of diffusion MRI models of white matter

Diffusion MRI is arguably the method of choice for characterizing white matter microstructure in vivo. Over the typical duration of diffusion encoding, the displacement of water molecules is conveniently on a length scale similar to that of the underlying cellular structures. Moreover, water molecules in white matter are largely compartmentalized which enables biologically-inspired compartmental diffusion models to characterize and quantify the true biological microstructure. A plethora of white matter models have been proposed. However, overparameterization and mathematical fitting complications encourage the introduction of simplifying assumptions that vary between different approaches. These choices impact the quantitative estimation of model parameters with potential detriments to their biological accuracy and promised specificity. First, we review biophysical white matter models in use and recapitulate their underlying assumptions and realms of applicability. Second, we present up-to-date efforts to validate parameters estimated from biophysical models. Simulations and dedicated phantoms are useful in assessing the performance of models when the ground truth is known. However, the biggest challenge remains the validation of the "biological accuracy" of estimated parameters. Complementary techniques such as microscopy of fixed tissue specimens have facilitated direct comparisons of estimates of white matter fiber orientation and densities. However, validation of compartmental diffusivities remains challenging, and complementary MRI-based techniques such as alternative diffusion encodings, compartment-specific contrast agents and metabolites have been used to validate diffusion models. Finally, white matter injury and disease pose additional challenges to modeling, which are also discussed. This review aims to provide an overview of the current state of models and their validation and to stimulate further research in the field to solve the remaining open questions and converge towards consensus.

Model selection for high b-value diffusion-weighted MRI of the prostate

PURPOSE

To assess the abilities of the standard mono-exponential (ME), bi-exponential (BE), diffusion kurtosis (DK) and stretched exponential (SE) models to characterize diffusion signal in malignant and prostatic tissues and determine which of the four models best characterizes these tissues on a per-voxel basis.

MATERIALS AND METHODS

This institutional-review-board-approved, HIPAA-compliant, retrospective study included 55 patients (median age, 61years; range, 42-77years) with untreated, biopsy-proven PCa who underwent endorectal coil MRI at 3-Tesla, diffusion-weighted MRI acquired at eight b-values from 0 to 2000s/mm². Estimated parameters were apparent diffusion coefficent (ME model); diffusion coefficients for the fast (D_fast) and slow (D_slow) components and fraction of fast component, f_fast (BE model); diffusion coefficient D, and kurtosis K (DK model); distributed diffusion coefficient DDC and α for (SE model). For one region-of-interest (ROI) in PZ and another in PCa in each patient, the corrected Akaike information criterion (AICc) and the Akaike weight (w) were calculated for each voxel.

RESULTS

Based on AICc and w, all non-monoexponential models outperformed the ME model in PZ and PCa. The DK model in PZ and SE model in PCa ROIs best fit the greatest average percentages of voxels (39% and 43%, respectively) and had the highest mean w (35±16×10^-2 and 41±22×10^-2, respectively).

CONCLUSION

DK and SE models best fit DWI data in PZ and PCa, and non-ME models consistently outperformed the ME model. Voxel-wise mapping of the preferential model demonstrated that the vast majority of voxels in either tissue type were best fit with one of the non-monoexponential models. At the given SNR levels, the maximum b-value of 2000s/mm² is not sufficiently high to identify the preferred non-monoexponential model.

Evaluation of fitting models for prostate tissue characterization using extended-range b-factor diffusion-weighted imaging

PURPOSE

To compare the fitting and tissue discrimination performance of biexponential, kurtosis, stretched exponential, and gamma distribution models for high b-factor diffusion-weighted images in prostate cancer.

METHODS

Diffusion-weighted images with 15 b-factors ranging from b = 0 to 3500 s/mm² were obtained in 62 prostate cancer patients. Pixel-wise signal decay fits for each model were evaluated with the Akaike Information Criterion (AIC). Parameter values for each model were determined within normal prostate and the index lesion. Their potential to differentiate normal from cancerous tissue was investigated through receiver operating characteristic analysis and comparison with Gleason score.

RESULTS

The biexponential slow diffusion fraction f_slow , the apparent kurtosis diffusion coefficient ADC_K , and the excess kurtosis factor K differ significantly among normal peripheral zone (PZ), normal transition zone (TZ), tumor PZ, and tumor TZ. Biexponential and gamma distribution models result in the lowest AIC, indicating a superior fit. Maximum areas under the curve (AUCs) of all models ranged from 0.93 to 0.96 for the PZ and from 0.95 to 0.97 for the TZ. Similar AUCs also result from the apparent diffusion coefficient (ADC) of a monoexponential fit to a b-factor sub-range up to 1250 s/mm² . For kurtosis and stretched exponential models, single parameters yield the highest AUCs, whereas for the biexponential and gamma distribution models, linear combinations of parameters produce the highest AUCs. Parameters with high AUC show a trend in differentiating low from high Gleason score, whereas parameters with low AUC show no such ability.

CONCLUSION

All models, including a monoexponential fit to a lower-b sub-range, achieve similar AUCs for discrimination of normal and cancer tissue. The biexponential model, which is favored statistically, also appears to provide insight into disease-related microstructural changes. Magn Reson Med 79:2346-2358, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Assessment of early treatment response to neoadjuvant chemotherapy in breast cancer using non-mono-exponential diffusion models: a feasibility study comparing the baseline and mid-treatment MRI examinations

OBJECTIVES

To assess the feasibility of the mono-exponential, bi-exponential and stretched-exponential models in evaluating response of breast tumours to neoadjuvant chemotherapy (NACT) at 3 T.

METHODS

Thirty-six female patients (median age 53, range 32-75 years) with invasive breast cancer undergoing NACT were enrolled for diffusion-weighted MRI (DW-MRI) prior to the start of treatment. For assessment of early response, changes in parameters were evaluated on mid-treatment MRI in 22 patients. DW-MRI was performed using eight b values (0, 30, 60, 90, 120, 300, 600, 900 s/mm2). Apparent diffusion coefficient (ADC), tissue diffusion coefficient (D t), vascular fraction (ƒ), distributed diffusion coefficient (DDC) and alpha (α) parameters were derived. Then t tests compared the baseline and changes in parameters between response groups. Repeatability was assessed at inter- and intraobserver levels.

RESULTS

All patients underwent baseline MRI whereas 22 lesions were available at mid-treatment. At pretreatment, mean diffusion coefficients demonstrated significant differences between groups (p < 0.05). At mid-treatment, percentage increase in ADC and DDC showed significant differences between responders (49 % and 43 %) and non-responders (21 % and 32 %) (p = 0.03, p = 0.04). Overall, stretched-exponential parameters showed excellent repeatability.

CONCLUSION

DW-MRI is sensitive to baseline and early treatment changes in breast cancer using non-mono-exponential models, and the stretched-exponential model can potentially monitor such changes.

KEY POINTS

• Baseline diffusion coefficients demonstrated significant differences between complete pathological responders and non-responders. • Increase in ADC and DDC at mid-treatment can discriminate responders and non-responders. • The ƒ fraction at mid-treatment decreased in responders whereas increased in non-responders. • The mono- and stretched-exponential models showed excellent inter- and intrarater repeatability. • Treatment effects can potentially be assessed by non-mono-exponential diffusion models.

Diffusion magnetic resonance imaging: A molecular imaging tool caught between hope, hype and the real world of “personalized oncology”

“Personalized oncology” is a multi-disciplinary science, which requires inputs from various streams for optimal patient management. Humongous progress in the treatment modalities available and the increasing need to provide functional information in addition to the morphological data; has led to leaping progress in the field of imaging. Magnetic resonance imaging has undergone tremendous progress with various newer MR techniques providing vital functional information and is becoming the cornerstone of “radiomics/radiogenomics”. Diffusion-weighted imaging is one such technique which capitalizes on the tendency of water protons to diffuse randomly in a given system. This technique has revolutionized oncological imaging, by giving vital qualitative and quantitative information regarding tumor biology which helps in detection, characterization and post treatment surveillance of the lesions and challenging the notion that “one size fits all”. It has been applied at various sites with different clinical experience. We hereby present a brief review of this novel functional imaging tool, with its application in “personalized oncology”.

Evaluation of different mathematical models and different b-value ranges of diffusion-weighted imaging in peripheral zone prostate cancer detection using b-value up to 4500 s/mm2

OBJECTIVES

To evaluate the diagnostic performance of different mathematical models and different b-value ranges of diffusion-weighted imaging (DWI) in peripheral zone prostate cancer (PZ PCa) detection.

METHODS

Fifty-six patients with histologically proven PZ PCa who underwent DWI-magnetic resonance imaging (MRI) using 21 b-values (0-4500 s/mm2) were included. The mean signal intensities of the regions of interest (ROIs) placed in benign PZs and cancerous tissues on DWI images were fitted using mono-exponential, bi-exponential, stretched-exponential, and kurtosis models. The b-values were divided into four ranges: 0-1000, 0-2000, 0-3200, and 0-4500 s/mm2, grouped as A, B, C, and D, respectively. ADC, <D>, D*, f, DDC, α, Dapp, and Kapp were estimated for each group. The adjusted coefficient of determination (R2) was calculated to measure goodness-of-fit. Receiver operating characteristic curve analysis was performed to evaluate the diagnostic performance of the parameters.

RESULTS

All parameters except D* showed significant differences between cancerous tissues and benign PZs in each group. The area under the curve values (AUCs) of ADC were comparable in groups C and D (p = 0.980) and were significantly higher than those in groups A and B (p< 0.05 for all). The AUCs of ADC and Kapp in groups B and C were similar (p = 0.07 and p = 0.954), and were significantly higher than the other parameters (p< 0.001 for all). The AUCs of ADC in group D was slightly higher than Kapp (p = 0.002), and both were significantly higher than the other parameters (p< 0.001 for all).

CONCLUSIONS

ADC derived from conventional mono-exponential high b-value (3200 s/mm2) models is an optimal parameter for PZ PCa detection.

Comparison of non-Gaussian and Gaussian diffusion models of diffusion weighted imaging of rectal cancer at 3.0 T MRI

Water molecular diffusion in vivo tissue is much more complicated. We aimed to compare non-Gaussian diffusion models of diffusion-weighted imaging (DWI) including intra-voxel incoherent motion (IVIM), stretched-exponential model (SEM) and Gaussian diffusion model at 3.0 T MRI in patients with rectal cancer, and to determine the optimal model for investigating the water diffusion properties and characterization of rectal carcinoma. Fifty-nine consecutive patients with pathologically confirmed rectal adenocarcinoma underwent DWI with 16 b-values at a 3.0 T MRI system. DWI signals were fitted to the mono-exponential and non-Gaussian diffusion models (IVIM-mono, IVIM-bi and SEM) on primary tumor and adjacent normal rectal tissue. Parameters of standard apparent diffusion coefficient (ADC), slow- and fast-ADC, fraction of fast ADC (f), α value and distributed diffusion coefficient (DDC) were generated and compared between the tumor and normal tissues. The SEM exhibited the best fitting results of actual DWI signal in rectal cancer and the normal rectal wall (R² = 0.998, 0.999 respectively). The DDC achieved relatively high area under the curve (AUC = 0.980) in differentiating tumor from normal rectal wall. Non-Gaussian diffusion models could assess tissue properties more accurately than the ADC derived Gaussian diffusion model. SEM may be used as a potential optimal model for characterization of rectal cancer.

Monoexponential, biexponential, and stretched exponential diffusion-weighted imaging models: Quantitative biomarkers for differentiating renal clear cell carcinoma and minimal fat angiomyolipoma

PURPOSE

To determine the utility of various diffusion parameters obtained from monoexponential, biexponential, and stretched exponential diffusion-weighted imaging (DWI) models in differentiating between minimal fat angiomyolipoma (MFAML) and clear cell renal cell carcinoma (ccRCC).

MATERIALS AND METHODS

One hundred thirty-one patients with pathologically confirmed MFAML (n = 27) or ccRCC (n = 104) underwent multi-b value DWI (0∼1700 s/mm² ) imaging at 3.0 Tesla MRI. An isotropic apparent diffusion coefficient (ADC) was calculated from diffusion-weighted images by using a monoexponential model. A pseudo-ADC (D_p ), true ADC (D_t ), and perfusion fraction (f_p ) were calculated from diffusion-weighted images by using a biexponential model. A water molecular diffusion heterogeneity index (α) and distributed diffusion coefficient (DDC) were calculated from diffusion-weighted images by using a stretched exponential model. All parameters were compared between MFAML and ccRCC by using the Student's t test. Receiver operating characteristic and intraclass correlation coefficient analysis were used for statistical evaluations.

RESULTS

ADC, D_t , and α values were significantly lower in the MFAML group than in the ccRCC group (P < 0.001). D_p , f_p , and DDC values were slightly higher in the MFAML group than in the ccRCC group; however, the difference was not significant (P = 0.136, 0.090, and 0.424, respectively). The AUC values for both α (0.953) and D_t (0.964) were significantly higher than those for ADC (0860), D_p (0.605), f_p (0.596), and DDC (0.477) in the differentiation of MFAML from ccRCC (P < 0.001).

CONCLUSION

Water molecular diffusion heterogeneity index (α) and D_t may provide additional information and could lead to improved differentiation with better sensitivity and specificity between MFAML and ccRCC compared with conventional diffusion parameters.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 2 J. MAGN. RESON. IMAGING 2017;46:240-247.

Multiparametric MR diffusion-weighted imaging for monitoring the ultra-early treatment effect of sorafenib in human hepatocellular carcinoma xenografts

PURPOSE

To investigate the value of multiparametric magnetic resonance imaging (MRI) diffusion-weighted imaging (DWI) for monitoring the ultra-early (within 24 hours) treatment effect of sorafenib in human hepatocellular carcinoma (HCC) xenografts.

MATERIALS AND METHODS

With institutional Animal Care and Use Committee approval, 16 BALB/c nude mice bearing subcutaneous HCC xenografts underwent serial Gaussian and non-Gaussian DWI at baseline and 1, 3, 6, 12, and 24 hours posttreatment using a 1.5T whole-body MRI system. Gaussian-DWI-derived apparent diffusion coefficient (ADC), D, D*, and f, and non-Gaussian-DWI-derived MD, MK, DDC, and α were calculated and compared between the control (n = 6) and sorafenib-treated groups (n = 10) with respect to each timepoint using Mann-Whitney or Wilcoxon signed-rank test. Results were validated by pathology.

RESULTS

Compared to baseline, ADC and D at 1 hour posttreatment (P = 0.005 and P = 0.013, respectively) and MD and DDC at 3 hours posttreatment (P = 0.009 and P = 0.005, respectively) significantly decreased and remained lower through 12 hours of follow-up (P = 0.005-0.022), followed by recovery to baseline levels at 24 hours posttreatment (P = 0.139-0.646). MK significantly increased at 1 hour posttreatment (P = 0.013) and remained higher through 24 hours of follow-up (P = 0.009-0.028). No significant differences were found in D*, f, and α at different timepoints (P = 0.188-0.714). Light microscopy did not reveal abnormal findings until 3 hours posttreatment, when central patchy necrosis was observed; more prominent diffuse necrosis was observed at 24 hours. Electron microscopy revealed swollen mitochondria at 1 hour posttreatment and accumulation of intracellular autophagosomes from 3 to 24 hours posttreatment.

CONCLUSION

Multiparametric DWI might evaluate therapeutic effects of sorafenib in HCC, where metrics of ADC, D, and MK could potentially serve as imaging biomarkers for monitoring therapeutic effects as early as 1 hour after treatment. Level of Evidence 1 Technical Efficacy: Stage 4 J. MAGN. RESON. IMAGING 2017;46:248-256.

Bench to bedside molecular functional imaging in translational cancer medicine: to image or to imagine?

Ongoing research on malignant and normal cell biology has substantially enhanced the understanding of the biology of cancer and carcinogenesis. This has led to the development of methods to image the evolution of cancer, target specific biological molecules, and study the anti-tumour effects of novel therapeutic agents. At the same time, there has been a paradigm shift in the field of oncological imaging from purely structural or functional imaging to combined multimodal structure-function approaches that enable the assessment of malignancy from all aspects (including molecular and functional level) in a single examination. The evolving molecular functional imaging using specific molecular targets (especially with combined positron-emission tomography [PET] computed tomography [CT] using 2- [(18)F]-fluoro-2-deoxy-D-glucose [FDG] and other novel PET tracers) has great potential in translational research, giving specific quantitative information with regard to tumour activity, and has been of pivotal importance in diagnoses and therapy tailoring. Furthermore, molecular functional imaging has taken a key place in the present era of translational cancer research, producing an important tool to study and evolve newer receptor-targeted therapies, gene therapies, and in cancer stem cell research, which could form the basis to translate these agents into clinical practice, popularly termed "theranostics". Targeted molecular imaging needs to be developed in close association with biotechnology, information technology, and basic translational scientists for its best utility. This article reviews the current role of molecular functional imaging as one of the main pillars of translational research.