Prediction of breast cancer and axillary positive-node response to neoadjuvant chemotherapy based on multi-parametric magnetic resonance imaging radiomics models

PURPOSE

Accurate identification of primary breast cancer and axillary positive-node response to neoadjuvant chemotherapy (NAC) is important for determining appropriate surgery strategies. We aimed to develop combining models based on breast multi-parametric magnetic resonance imaging and clinicopathologic characteristics for predicting therapeutic response of primary tumor and axillary positive-node prior to treatment.

MATERIALS AND METHODS

A total of 268 breast cancer patients who completed NAC and underwent surgery were enrolled. Radiomics features and clinicopathologic characteristics were analyzed through the analysis of variance and the least absolute shrinkage and selection operator algorithm. Finally, 24 and 28 optimal features were selected to construct machine learning models based on 6 algorithms for predicting each clinical outcome, respectively. The diagnostic performances of models were evaluated in the testing set by the area under the curve (AUC), sensitivity, specificity, and accuracy.

RESULTS

Of the 268 patients, 94 (35.1 %) achieved breast cancer pathological complete response (bpCR) and of the 240 patients with clinical positive-node, 120 (50.0 %) achieved axillary lymph node pathological complete response (apCR). The multi-layer perception (MLP) algorithm yielded the best diagnostic performances in predicting apCR with an AUC of 0.825 (95 % CI, 0.764-0.886) and an accuracy of 77.1 %. And MLP also outperformed other models in predicting bpCR with an AUC of 0.852 (95 % CI, 0.798-0.906) and an accuracy of 81.3 %.

CONCLUSIONS

Our study established non-invasive combining models to predict the therapeutic response of primary breast cancer and axillary positive-node prior to NAC, which may help to modify preoperative treatment and determine post-NAC surgery strategy.

Fractal analysis improves tumour size measurement on computed tomography in pancreatic ductal adenocarcinoma: comparison with gross pathology and multi-parametric MRI

Objectives

Tumour size measurement is pivotal for staging and stratifying patients with pancreatic ductal adenocarcinoma (PDA). However, computed tomography (CT) frequently underestimates tumour size due to insufficient depiction of the tumour rim. CT-derived fractal dimension (FD) maps might help to visualise perfusion chaos, thus allowing more realistic size measurement.

Methods

In 46 patients with histology-proven PDA, we compared tumour size measurements in routine multiphasic CT scans, CT-derived FD maps, multi-parametric magnetic resonance imaging (mpMRI), and, where available, gross pathology of resected specimens. Gross pathology was available as reference for diameter measurement in a discovery cohort of 10 patients. The remaining 36 patients constituted a separate validation cohort with mpMRI as reference for diameter and volume.

Results

Median RECIST diameter of all included tumours was 40 mm (range: 18–82 mm). In the discovery cohort, we found significant (p = 0.03) underestimation of tumour diameter on CT compared with gross pathology (Δdiameter_3D = −5.7 mm), while realistic diameter measurements were obtained from FD maps (Δdiameter_3D = 0.6 mm) and mpMRI (Δdiameter_3D = −0.9 mm), with excellent correlation between the two (R² = 0.88). In the validation cohort, CT also systematically underestimated tumour size in comparison to mpMRI (Δdiameter_3D = −10.6 mm, Δvolume = −10.2 mL), especially in larger tumours. In contrast, FD map measurements agreed excellently with mpMRI (Δdiameter_3D = +1.5 mm, Δvolume = −0.6 mL). Quantitative perfusion chaos was significantly (p = 0.001) higher in the tumour rim (FD_rim = 4.43) compared to the core (FD_core = 4.37) and remote pancreas (FD_pancreas = 4.28).

Conclusions

In PDA, fractal analysis visualises perfusion chaos in the tumour rim and improves size measurement on CT in comparison to gross pathology and mpMRI, thus compensating for size underestimation from routine CT.

Key Points

• CT-based measurement of tumour size in pancreatic adenocarcinoma systematically underestimates both tumour diameter (Δdiameter = −10.6 mm) and volume (Δvolume = −10.2 mL), especially in larger tumours.

• Fractal analysis provides maps of the fractal dimension (FD), which enable a more reliable and size-independent measurement using gross pathology or multi-parametric MRI as reference standards.

• FD quantifies perfusion chaos—the underlying pathophysiological principle—and can separate the more chaotic tumour rim from the tumour core and adjacent non-tumourous pancreas tissue.

Histogram analysis of diffusion-weighted magnetic resonance imaging as a biomarker to predict LNM in T3 stage rectal carcinoma

BACKGROUND

Preoperative identification of rectal cancer lymph node status is crucial for patient prognosis and treatment decisions. Rectal magnetic resonance imaging (MRI) plays an essential role in the preoperative staging of rectal cancer, but its ability to predict lymph node metastasis (LNM) is insufficient. This study explored the value of histogram features of primary lesions on multi-parametric MRI for predicting LNM of stage T3 rectal carcinoma.

METHODS

We retrospectively analyzed 175 patients with stage T3 rectal cancer who underwent preoperative MRI, including diffusion-weighted imaging (DWI) before surgery. 62 patients were included in the LNM group, and 113 patients were included in the non-LNM group. Texture features were calculated from histograms derived from T2 weighted imaging (T2WI), DWI, ADC, and T2 maps. Stepwise logistic regression analysis was used to screen independent predictors of LNM from clinical features, imaging features, and histogram features. Predictive performance was evaluated by receiver operating characteristic (ROC) curve analysis. Finally, a nomogram was established for predicting the risk of LNM.

RESULTS

The clinical, imaging and histogram features were analyzed by stepwise logistic regression. Preoperative carbohydrate antigen 199 level (p = 0.009), MRN stage (p < 0.001), _T2WIKurtosis (p = 0.010), _DWIMode (p = 0.038), _DWICV (p = 0.038), and _T2-mapP5 (p = 0.007) were independent predictors of LNM. These factors were combined to form the best predictive model. The model reached an area under the ROC curve (AUC) of 0.860, with a sensitivity of 72.8% and a specificity of 85.5%.

CONCLUSION

The histogram features on multi-parametric MRI of the primary tumor in rectal cancer were related to LN status, which is helpful for improving the ability to predict LNM of stage T3 rectal cancer.

Diagnosis of focal liver lesions with deep learning-based multi-channel analysis of hepatocyte-specific contrast-enhanced magnetic resonance imaging

BACKGROUND

The nature of input data is an essential factor when training neural networks. Research concerning magnetic resonance imaging (MRI)-based diagnosis of liver tumors using deep learning has been rapidly advancing. Still, evidence to support the utilization of multi-dimensional and multi-parametric image data is lacking. Due to higher information content, three-dimensional input should presumably result in higher classification precision. Also, the differentiation between focal liver lesions (FLLs) can only be plausible with simultaneous analysis of multi-sequence MRI images.

AIM

To compare diagnostic efficiency of two-dimensional (2D) and three-dimensional (3D)-densely connected convolutional neural networks (DenseNet) for FLLs on multi-sequence MRI.

METHODS

We retrospectively collected T2-weighted, gadoxetate disodium-enhanced arterial phase, portal venous phase, and hepatobiliary phase MRI scans from patients with focal nodular hyperplasia (FNH), hepatocellular carcinomas (HCC) or liver metastases (MET). Our search identified 71 FNH, 69 HCC and 76 MET. After volume registration, the same three most representative axial slices from all sequences were combined into four-channel images to train the 2D-DenseNet264 network. Identical bounding boxes were selected on all scans and stacked into 4D volumes to train the 3D-DenseNet264 model. The test set consisted of 10-10-10 tumors. The performance of the models was compared using area under the receiver operating characteristic curve (AUROC), specificity, sensitivity, positive predictive values (PPV), negative predictive values (NPV), and f1 scores.

RESULTS

The average AUC value of the 2D model (0.98) was slightly higher than that of the 3D model (0.94). Mean PPV, sensitivity, NPV, specificity and f1 scores (0.94, 0.93, 0.97, 0.97, and 0.93) of the 2D model were also superior to metrics of the 3D model (0.84, 0.83, 0.92, 0.92, and 0.83). The classification metrics of FNH were 0.91, 1.00, 1.00, 0.95, and 0.95 using the 2D and 0.90, 0.90, 0.95, 0.95, and 0.90 using the 3D models. The 2D and 3D networks' performance in the diagnosis of HCC were 1.00, 0.80, 0.91, 1.00, and 0.89 and 0.88, 0.70, 0.86, 0.95, and 0.78, respectively; while the evaluation of MET lesions resulted in 0.91, 1.00, 1.00, 0.95, and 0.95 and 0.75, 0.90, 0.94, 0.85, and 0.82 using the 2D and 3D networks, respectively.

CONCLUSION

Both 2D and 3D-DenseNets can differentiate FNH, HCC and MET with good accuracy when trained on hepatocyte-specific contrast-enhanced multi-sequence MRI volumes.

Quantitative MRI using STrategically Acquired Gradient Echo (STAGE): optimization for 1.5 T scanners and T1 relaxation map validation

OBJECTIVES

The strategically acquired gradient echo (STAGE) protocol, developed for 3T scanners, allows one to derive quantitative maps such as T1, T2*, proton density, and quantitative susceptibility mapping in about 5 min. Our aim was to adapt the STAGE sequences for 1.5T scanners which are still commonly used in clinical practice. Furthermore, the accuracy and repeatability of the STAGE-derived T1 estimate were tested.

METHODS

Flip angle (FA) optimization was performed using a theoretical simulation by maximizing signal-to-noise ratio, contrast-to-noise ratio, and T1 precision. The FA choice was further refined with the ISMRM/NIST phantom and in vivo acquisitions. The accuracy of the T1 estimate was assessed by comparing STAGE-derived T1 values with T1 maps obtained with an inversion recovery sequence. T1 accuracy was investigated for both the phantom and in vivo data. Finally, one subject was acquired 10 times once a week and a group of 27 subjects was scanned once. The T1 coefficient of variation (COV) was computed to assess scan-rescan and physiological variability, respectively.

RESULTS

The FA_1,2 = 7°,38° were identified as the optimal FA pair at 1.5T. The T1 estimate errors were below 3% and 5% for phantom and in vivo measurements, respectively. COV for different tissues ranged from 1.8 to 4.8% for physiological variability, and between 0.8 and 2% for scan-rescan repeatability.

CONCLUSION

The optimized STAGE protocol can provide accurate and repeatable T1 mapping along with other qualitative images and quantitative maps in about 7 min on 1.5T scanners. This study provides the groundwork to assess the role of STAGE in clinical settings.

KEY POINTS

• The STAGE imaging protocol was optimized for use on 1.5T field strength scanners. • A practical STAGE protocol makes it possible to derive quantitative maps (i.e., T1, T2*, PD, and QSM) in about 7 min at 1.5T. • The T1 estimate derived from the STAGE protocol showed good accuracy and repeatability.

Advanced zoomed diffusion-weighted imaging vs. full-field-of-view diffusion-weighted imaging in prostate cancer detection: a radiomic features study

OBJECTIVES

We aimed to compare the efficiency of prostate cancer (PCa) detection using a radiomics signature based on advanced zoomed diffusion-weighted imaging and conventional full-field-of-view DWI.

METHODS

A total of 136 patients, including 73 patients with PCa and 63 without PCa, underwent multi-parametric magnetic resonance imaging (mp-MRI). Radiomic features were extracted from prostate lesion areas segmented on full-field-of-view DWI with b-value = 1500 s/mm² (f-DWI_b1500), advanced zoomed DWI images with b-value = 1500 s/mm² (z-DWI_b1500), calculated zoomed DWI with b-value = 2000 s/mm² (z-calDWI_b2000), and apparent diffusion coefficient (ADC) maps derived from both sequences (f-ADC and z-ADC). Single-imaging modality radiomics signature, mp-MRI radiomics signature, and a mixed model based on mp-MRI and clinically independent risk factors were built to predict PCa probability. The diagnostic efficacy and the potential net benefits of each model were evaluated.

RESULTS

Both z-DWI_b1500 and z-calDWI_b2000 had significantly better predictive performance than f-DWIb1500 (z-DWIb1500 vs. f-DWIb1500: p = 0.048; z-calDWIb2000 vs. f-DWIb1500: p = 0.014). z-ADC had a slightly higher area under the curve (AUC) value compared with f-ADC value but was not significantly different (p = 0.127). For predicting the presence of PCa, the AUCs of clinical independent risk factors model, mp-MRI model, and mixed model were 0.81, 0.93, and 0.94 in training sets, and 0.74, 0.92, and 0.93 in validation sets, respectively.

CONCLUSION

Radiomics signatures based on the z-DWI technology had better diagnostic accuracy for PCa than that based on the f-DWI technology. The mixed model was better at diagnosing PCa and guiding clinical interventions for patients with suspected PCa compared with mp-MRI signatures and clinically independent risk factors.

KEY POINTS

• Advanced zoomed DWI technology can improve the diagnostic accuracy of radiomics signatures for PCa. • Radiomics signatures based on z-calDWI_b2000 have the best diagnostic performance among individual imaging modalities. • Compared with the independent clinical risk factors and the mp-MRI model, the mixed model has the best diagnostic efficiency.

Application of hierarchical clustering to multi-parametric MR in prostate: Differentiation of tumor and normal tissue with high accuracy

PURPOSE

Hierarchical clustering (HC), an unsupervised machine learning (ML) technique, was applied to multi-parametric MR (mp-MR) for prostate cancer (PCa). The aim of this study is to demonstrate HC can diagnose PCa in a straightforward interpretable way, in contrast to deep learning (DL) techniques.

METHODS

HC was constructed using mp-MR including intravoxel incoherent motion, diffusion kurtosis imaging, and dynamic contrast-enhanced MRI from 40 tumor and normal tissues in peripheral zone (PZ) and 23 tumor and normal tissues in transition zone (TZ). HC model was optimized by assessing the combinations of several dissimilarity and linkage methods. Goodness of HC model was validated by internal methods.

RESULTS

Accuracy for differentiating tumor and normal tissue by optimal HC model was 96.3% in PZ and 97.8% in TZ, comparable to current clinical standards. Relationship between input (DWI and permeability parameters) and output (tumor and normal tissue cluster) was shown by heat maps, consistent with literature.

CONCLUSION

HC can accurately differentiate PCa and normal tissue, comparable to state-of-the-art diffusion based parameters. Contrary to DL techniques, HC is an operator-independent ML technique producing results that can be interpreted such that the results can be knowledgeably judged.

Clinically significant prostate cancer detection and segmentation in low-risk patients using a convolutional neural network on multi-parametric MRI

Objectives

To develop an automatic method for identification and segmentation of clinically significant prostate cancer in low-risk patients and to evaluate the performance in a routine clinical setting.

Methods

A consecutive cohort (n = 292) from a prospective database of low-risk patients eligible for the active surveillance was selected. A 3-T multi-parametric MRI at 3 months after inclusion was performed. Histopathology from biopsies was used as reference standard. MRI positivity was defined as PI-RADS score ≥ 3, histopathology positivity was defined as ISUP grade ≥ 2. The selected cohort contained four patient groups: (1) MRI-positive targeted biopsy-positive (n = 116), (2) MRI-negative systematic biopsy-negative (n = 55), (3) MRI-positive targeted biopsy-negative (n = 113), (4) MRI-negative systematic biopsy-positive (n = 8). Group 1 was further divided into three sets and a 3D convolutional neural network was trained using different combinations of these sets. Two MRI sequences (T2w, b = 800 DWI) and the ADC map were used as separate input channels for the model. After training, the model was evaluated on the remaining group 1 patients together with the patients of groups 2 and 3 to identify and segment clinically significant prostate cancer.

Results

The average sensitivity achieved was 82–92% at an average specificity of 43–76% with an area under the curve (AUC) of 0.65 to 0.89 for different lesion volumes ranging from > 0.03 to > 0.5 cc.

Conclusions

The proposed deep learning computer-aided method yields promising results in identification and segmentation of clinically significant prostate cancer and in confirming low-risk cancer (ISUP grade ≤ 1) in patients on active surveillance.

Key Points

• Clinically significant prostate cancer identification and segmentation on multi-parametric MRI is feasible in low-risk patients using a deep neural network.

• The deep neural network for significant prostate cancer localization performs better for lesions with larger volumes sizes (> 0.5 cc) as compared to small lesions (> 0.03 cc).

• For the evaluation of automatic prostate cancer segmentation methods in the active surveillance cohort, the large discordance group (MRI positive, targeted biopsy negative) should be included.

Electronic supplementary material

The online version of this article (10.1007/s00330-020-07008-z) contains supplementary material, which is available to authorized users.

Multi-parametric qualitative and quantitative MRI assessment as predictor of histological grading in previously treated meningiomas

PURPOSE

Meningiomas are mainly benign tumors, though a considerable proportion shows aggressive behaviors histologically consistent with atypia/anaplasia. Histopathological grading is usually assessed through invasive procedures, which is not always feasible due to the inaccessibility of the lesion or to treatment contraindications. Therefore, we propose a multi-parametric MRI assessment as a predictor of meningioma histopathological grading.

METHODS

Seventy-three patients with 74 histologically proven and previously treated meningiomas were retrospectively enrolled (42 WHO I, 24 WHO II, 8 WHO III) and studied with MRI including T2 TSE, FLAIR, Gradient Echo, DWI, and pre- and post-contrast T1 sequences. Lesion masks were segmented on post-contrast T1 sequences and rigidly registered to ADC maps to extract quantitative parameters from conventional DWI and intravoxel incoherent motion model assessing tumor perfusion. Two expert neuroradiologists assessed morphological features of meningiomas with semi-quantitative scores.

RESULTS

Univariate analysis showed different distributions (p < 0.05) of quantitative diffusion parameters (Wilcoxon rank-sum test) and morphological features (Pearson's chi-square; Fisher's exact test) among meningiomas grouped in low-grade (WHO I) and higher grade forms (WHO II/III); the only exception consisted of the tumor-brain interface. A multivariate logistic regression, combining all parameters showing statistical significance in the univariate analysis, allowed discrimination between the groups of meningiomas with high sensitivity (0.968) and specificity (0.925). Heterogeneous contrast enhancement and low ADC were the best independent predictors of atypia and anaplasia.

CONCLUSION

Our multi-parametric MRI assessment showed high sensitivity and specificity in predicting histological grading of meningiomas. Such an assessment may be clinically useful in characterizing lesions without histological diagnosis. Key points • When surgery and biopsy are not feasible, parameters obtained from both conventional and diffusion-weighted MRI can predict atypia and anaplasia in meningiomas with high sensitivity and specificity. • Low ADC values and heterogeneous contrast enhancement are the best predictors of higher grade meningioma.

ESUR/ESUI consensus statements on multi-parametric MRI for the detection of clinically significant prostate cancer: quality requirements for image acquisition, interpretation and radiologists' training

OBJECTIVES

This study aims to define consensus-based criteria for acquiring and reporting prostate MRI and establishing prerequisites for image quality.

METHODS

A total of 44 leading urologists and urogenital radiologists who are experts in prostate cancer imaging from the European Society of Urogenital Radiology (ESUR) and EAU Section of Urologic Imaging (ESUI) participated in a Delphi consensus process. Panellists completed two rounds of questionnaires with 55 items under three headings: image quality assessment, interpretation and reporting, and radiologists' experience plus training centres. Of 55 questions, 31 were rated for agreement on a 9-point scale, and 24 were multiple-choice or open. For agreement items, there was consensus agreement with an agreement ≥ 70% (score 7-9) and disagreement of ≤ 15% of the panellists. For the other questions, a consensus was considered with ≥ 50% of votes.

RESULTS

Twenty-four out of 31 of agreement items and 11/16 of other questions reached consensus. Agreement statements were (1) reporting of image quality should be performed and implemented into clinical practice; (2) for interpretation performance, radiologists should use self-performance tests with histopathology feedback, compare their interpretation with expert-reading and use external performance assessments; and (3) radiologists must attend theoretical and hands-on courses before interpreting prostate MRI. Limitations are that the results are expert opinions and not based on systematic reviews or meta-analyses. There was no consensus on outcomes statements of prostate MRI assessment as quality marker.

CONCLUSIONS

An ESUR and ESUI expert panel showed high agreement (74%) on issues improving prostate MRI quality. Checking and reporting of image quality are mandatory. Prostate radiologists should attend theoretical and hands-on courses, followed by supervised education, and must perform regular performance assessments.

KEY POINTS

• Multi-parametric MRI in the diagnostic pathway of prostate cancer has a well-established upfront role in the recently updated European Association of Urology guideline and American Urological Association recommendations. • Suboptimal image acquisition and reporting at an individual level will result in clinicians losing confidence in the technique and returning to the (non-MRI) systematic biopsy pathway. Therefore, it is crucial to establish quality criteria for the acquisition and reporting of mpMRI. • To ensure high-quality prostate MRI, experts consider checking and reporting of image quality mandatory. Prostate radiologists must attend theoretical and hands-on courses, followed by supervised education, and must perform regular self- and external performance assessments.

Current status and future prospective of focal therapy for localized prostate cancer: development of multiparametric MRI, MRI-TRUS fusion image-guided biopsy, and treatment modalities

Multiparametric magnetic resonance imaging (mpMRI) has been increasingly used to diagnose clinically significant prostate cancer (csPC) because of its usefulness in combination with anatomic and functional data. MRI-targeted biopsy, such as MRI-transrectal ultrasound (TRUS) fusion image-guided prostate biopsy, has high accuracy in the detection and localization of csPC. This novel diagnostic technique contributes to the development of tailor-made medicine as focal therapy, which cures the csPC while preserving the anatomical structures related to urinary and sexual function. In the early days of focal therapy, TRUS-guided systematic biopsy was used for patient selection, and treatment was performed for patients with low-risk PC. With the introduction of mpMRI and mapping biopsy, the treatment range is now determined based on individualized cancer localization. In recent prospective studies, 87.4% of treated patients had intermediate- and high-risk PC. However, focal therapy has two main limitations. First, a randomized controlled trial would be difficult to design because of the differences in pathological features between patients undergoing focal therapy and radical treatment. Therefore, pair-matched studies and/or historical controlled studies have been performed to compare focal therapy and radical treatment. Second, no long-term (≥ 10-year) follow-up study has been performed. However, recent prospective studies have encouraged the use of focal therapy as a treatment strategy for localized PC because it contributes to high preservation of continence and erectile function.

Targeted and Systematic Biopsy for Diagnosis and Management of Prostate Cancer

The value of multi-parametric magnetic resonance imaging in the detection of clinically-significant prostate cancer is increasingly well-established, and has been adopted in current diagnostic pathways and clinical guidelines. Concurrently, the role of conventional ultrasound-guided systematic prostate biopsy is increasingly questioned. In this brief review, we evaluate the continued value of systematic biopsy including a review of prospective studies on targeted and systemic biopsies in the same patients. We also address current limitations of multi-parametric magnetic resonance imaging of the prostate.

STrategically Acquired Gradient Echo (STAGE) imaging, part III: Technical advances and clinical applications of a rapid multi-contrast multi-parametric brain imaging method

One major thrust in radiology today is image standardization with a focus on rapidly acquired quantitative multi-contrast information. This is critical for multi-center trials, for the collection of big data and for the use of artificial intelligence in evaluating the data. Strategically acquired gradient echo (STAGE) imaging is one such method that can provide 8 qualitative and 7 quantitative pieces of information in 5 min or less at 3 T. STAGE provides qualitative images in the form of proton density weighted images, T1 weighted images, T2* weighted images and simulated double inversion recovery (DIR) images. STAGE also provides quantitative data in the form of proton spin density, T1, T2* and susceptibility maps as well as segmentation of white matter, gray matter and cerebrospinal fluid. STAGE uses vendors' product gradient echo sequences. It can be applied from 0.35 T to 7 T across all manufacturers producing similar results in contrast and quantification of the data. In this paper, we discuss the strengths and weaknesses of STAGE, demonstrate its contrast-to-noise (CNR) behavior relative to a large clinical data set and introduce a few new image contrasts derived from STAGE, including DIR images and a new concept referred to as true susceptibility weighted imaging (tSWI) linked to fluid attenuated inversion recovery (FLAIR) or tSWI-FLAIR for the evaluation of multiple sclerosis lesions. The robustness of STAGE T1 mapping was tested using the NIST/NIH phantom, while the reproducibility was tested by scanning a given individual ten times in one session and the same subject scanned once a week over a 12-week period. Assessment of the CNR for the enhanced T1W image (T1WE) showed a significantly better contrast between gray matter and white matter than conventional T1W images in both patients with Parkinson's disease and healthy controls. We also present some clinical cases using STAGE imaging in patients with stroke, metastasis, multiple sclerosis and a fetus with ventriculomegaly. Overall, STAGE is a comprehensive protocol that provides the clinician with numerous qualitative and quantitative images.

Machine-learning in grading of gliomas based on multi-parametric magnetic resonance imaging at 3T

OBJECTIVE

The objective of this study was to assess the contribution of multi-parametric (mp) magnetic resonance imaging (MRI) quantitative features in the machine learning-based grading of gliomas with a multi-region-of-interests approach.

MATERIALS AND METHODS

Forty-three patients who were newly diagnosed as having a glioma were included in this study. The patients were scanned prior to any therapy using a standard brain tumor magnetic resonance (MR) imaging protocol that included T1 and T2-weighted, diffusion-weighted, diffusion tensor, MR perfusion and MR spectroscopic imaging. Three different regions-of-interest were drawn for each subject to encompass tumor, immediate tumor periphery, and distant peritumoral edema/normal. The normalized mp-MRI features were used to build machine-learning models for differentiating low-grade gliomas (WHO grades I and II) from high grades (WHO grades III and IV). In order to assess the contribution of regional mp-MRI quantitative features to the classification models, a support vector machine-based recursive feature elimination method was applied prior to classification.

RESULTS

A machine-learning model based on support vector machine algorithm with linear kernel achieved an accuracy of 93.0%, a specificity of 86.7%, and a sensitivity of 96.4% for the grading of gliomas using ten-fold cross validation based on the proposed subset of the mp-MRI features.

CONCLUSION

In this study, machine-learning based on multiregional and multi-parametric MRI data has proven to be an important tool in grading glial tumors accurately even in this limited patient population. Future studies are needed to investigate the use of machine learning algorithms for brain tumor classification in a larger patient cohort.

Integrated 68Gallium Labelled Prostate-Specific Membrane Antigen-11 Positron Emission Tomography/Magnetic Resonance Imaging Enhances Discriminatory Power of Multi-Parametric Prostate Magnetic Resonance Imaging

PURPOSE

To evaluate diagnostic accuracy of integrated 68Gallium labelled prostate-specific membrane antigen (68Ga-PSMA)-11 positron emission tomography (PET)/MRI in patients with primary prostate cancer (PCa) as compared to multi-parametric MRI.

MATERIAL AND METHODS

A total of 22 patients with recently diagnosed primary PCa underwent clinically indicated 68Ga-PSMA-11 PET/CT for initial staging followed by integrated 68Ga-PSMA-11 PET/MRI. Images of multi-parametric magnetic resonance imaging (mpMRI), PET and PET/MRI were evaluated separately by applying Prostate Imaging Reporting and Data System (PIRADSv2) for mpMRI and a 5-point Likert scale for PET and PET/MRI. Results were compared with pathology reports of biopsy or resection. Statistical analyses including receiver operating characteristics analysis were performed to compare the diagnostic performance of mpMRI, PET and PET/MRI.

RESULTS

PET and integrated PET/MRI demonstrated a higher diagnostic accuracy than mpMRI (area under the curve: mpMRI: 0.679, PET and PET/MRI: 0.951). The proportion of equivocal results (PIRADS 3 and Likert 3) was considerably higher in mpMRI than in PET and PET/MRI. In a notable proportion of equivocal PIRADS results, PET led to a correct shift towards higher suspicion of malignancy and enabled correct lesion classification.

CONCLUSION

Integrated 68Ga-PSMA-11 PET/MRI demonstrates higher diagnostic accuracy than mpMRI and is particularly valuable in tumours with equivocal results from PIRADS classification.

Robust and efficient pharmacokinetic parameter non-linear least squares estimation for dynamic contrast enhanced MRI of the prostate

PURPOSE

To describe an efficient numerical optimization technique using non-linear least squares to estimate perfusion parameters for the Tofts and extended Tofts models from dynamic contrast enhanced (DCE) MRI data and apply the technique to prostate cancer.

METHODS

Parameters were estimated by fitting the two Tofts-based perfusion models to the acquired data via non-linear least squares. We apply Variable Projection (VP) to convert the fitting problem from a multi-dimensional to a one-dimensional line search to improve computational efficiency and robustness. Using simulation and DCE-MRI studies in twenty patients with suspected prostate cancer, the VP-based solver was compared against the traditional Levenberg-Marquardt (LM) strategy for accuracy, noise amplification, robustness to converge, and computation time.

RESULTS

The simulation demonstrated that VP and LM were both accurate in that the medians closely matched assumed values across typical signal to noise ratio (SNR) levels for both Tofts models. VP and LM showed similar noise sensitivity. Studies using the patient data showed that the VP method reliably converged and matched results from LM with approximate 3× and 2× reductions in computation time for the standard (two-parameter) and extended (three-parameter) Tofts models. While LM failed to converge in 14% of the patient data, VP converged in the ideal 100%.

CONCLUSION

The VP-based method for non-linear least squares estimation of perfusion parameters for prostate MRI is equivalent in accuracy and robustness to noise, while being more reliably (100%) convergent and computationally about 3× (TM) and 2× (ETM) faster than the LM-based method.

A single mitochondrial DNA deletion accurately detects significant prostate cancer in men in the PSA 'grey zone'

Purpose

To determine the clinical performance of a blood-based test for clinically significant (CS) prostate cancer (PCa) (grade group ≥ 2) intended for use in men with prostate serum antigen levels in the ‘grey zone’ (PSA < 10 ng/ml). The test quantifies a previously described 3.4 kb mitochondrial DNA (mtDNA) deletion.

Methods

In a first prospective study of an MRI-guided re-biopsy population (n = 126), the 3.4 kb deletion and 18S rRNA gene were amplified from plasma. A diagnostic threshold was selected from the coordinates of the receiver operating characteristic curve and tested in a second population of men who were (n = 92) biopsy naïve when the mtDNA deletion was assayed and for whom those diagnosed with cancer on initial biopsy were treated with radical prostatectomy.

Results

The 3.4 kb deletion was a good predictor of CS PCa in the image-guided re-biopsy population [AUC 0.84, (95% CI 0.73–0.95)] and the selected threshold corresponded to a sensitivity of 87% [95% CI, 70–96%], specificity of 68% [95% CI, 47–85%] and negative predictive value (NPV) of 97%. Applying this threshold to the second population showed this deletion to be a strong predictor of CS cancer [AUC 0.98, (95% CI 0.94–1.02)], independent of PSA or age [sensitivity 100% (95% CI, 93–100%), specificity 90% (95%CI 73–98%) and NPV 100%].

Conclusion

The 3.4 kb deletion in plasma is an accurate predictor of CS cancer for men in the PSA ‘grey zone’. Used in advance of biopsy for improved patient selection, this deletion may reduce the number of biopsies needed to diagnose CS prostate cancers.

Electronic supplementary material

The online version of this article (10.1007/s00345-017-2152-z) contains supplementary material, which is available to authorized users.

Multi-parametric MR imaging of transition zone prostate cancer: Imaging features, detection and staging

Magnetic resonance (MR) imaging has been increasingly used in the evaluation of prostate cancer. As studies have suggested that the majority of cancers arise from the peripheral zone (PZ), MR imaging has focused on the PZ of the prostate gland thus far. However, a considerable number of cancers (up to 30%) originate in the transition zone (TZ), substantially contributing to morbidity and mortality. Therefore, research is needed on the TZ of the prostate gland. Recently, MR imaging and advanced MR techniques have been gaining acceptance in evaluation of the TZ. In this article, the MR imaging features of TZ prostate cancers, the role of MR imaging in TZ cancer detection and staging, and recent advanced MR techniques will be discussed in light of the literature.