Preoperative Differentiation of Uterine Sarcoma from Leiomyoma: Comparison of Three Models Based on Different Segmentation Volumes Using Radiomics

Application of convolutional neural network for differentiating ovarian thecoma-fibroma and solid ovarian cancer based on MRI

BACKGROUND

Ovarian thecoma-fibroma and solid ovarian cancer have similar clinical and imaging features, and it is difficult for radiologists to differentiate them. Since the treatment and prognosis of them are different, accurate characterization is crucial.

PURPOSE

To non-invasively differentiate ovarian thecoma-fibroma and solid ovarian cancer by convolutional neural network based on magnetic resonance imaging (MRI), and to provide the interpretability of the model.

MATERIAL AND METHODS

A total of 156 tumors, including 86 ovarian thecoma-fibroma and 70 solid ovarian cancer, were split into the training set, the validation set, and the test set according to the ratio of 8:1:1 by stratified random sampling. In this study, we used four different networks, two different weight modes, two different optimizers, and four different sizes of regions of interest (ROI) to test the model performance. This process was repeated 10 times to calculate the average performance of the test set. The gradient weighted class activation mapping (Grad-CAM) was used to explain how the model makes classification decisions by visual location map.

RESULTS

ResNet18, which had pre-trained weight, using Adam and one multiple ROI circumscribed rectangle, achieved best performance. The average accuracy, precision, recall, and AUC were 0.852, 0.828, 0.848, and 0.919 (P < 0.01), respectively. Grad-CAM showed areas associated with classification appeared on the edge or interior of ovarian thecoma-fibroma and the interior of solid ovarian cancer.

CONCLUSION

This study shows that convolution neural network based on MRI can be helpful for radiologists in differentiating ovarian thecoma-fibroma and solid ovarian cancer.

MRI, clinical, and radiomic models for differentiation of uterine leiomyosarcoma and leiomyoma

PURPOSE

To assess the predictive ability of conventional MRI features and MRI texture features in differentiating uterine leiomyoma (LM) from uterine leiomyosarcoma (LMS).

METHODS

This single-center, IRB-approved, HIPAA-compliant retrospective study included 108 patients (69 LM, 39 LMS) who had pathology, preoperative MRI, and clinical data available at our tertiary academic institution. Two radiologists independently evaluated 14 features on preoperative MRI. Texture features based on 3D segmentation were extracted from T2W-weighted MRI (T2WI) using commercially available texture software (TexRAD™, Feedback Medical Ltd., Great Britain). MRI conventional features, and clinical and MRI texture features were compared between LM and LMS groups. Dataset was randomly divided into training (86 cases) and testing (22 cases) cohorts (8:2 ratio); training cohort was further subdivided into training and validation sets using ten-fold cross-validation. Optimal radiomics model was selected out of 90 different machine learning pipelines and five models containing different combinations of MRI, clinical, and radiomics variables.

RESULTS

12/14 MRI conventional features and 2/2 clinical features were significantly different between LM and LMS groups. MRI conventional features had moderate to excellent inter-reader agreement for all but two features. Models combining MRI conventional and clinical features (AUC 0.956) and MRI conventional, clinical, and radiomics features (AUC 0.989) had better performance compared to models containing MRI conventional features alone (AUC 0.846 and 0.890) or radiomics features alone (0.929).

CONCLUSION

While multiple MRI and clinical features differed between LM and LMS groups, the model combining MRI, clinical, and radiomic features had the best predictive ability but was only marginally better than a model utilizing conventional MRI and clinical data alone.

Differentiating Uterine Sarcoma From Atypical Leiomyoma on Preoperative Magnetic Resonance Imaging Using Logistic Regression Classifier: Added Value of Diffusion-Weighted Imaging-Based Quantitative Parameters

OBJECTIVE

To evaluate the added value of diffusion-weighted imaging (DWI)-based quantitative parameters to distinguish uterine sarcomas from atypical leiomyomas on preoperative magnetic resonance imaging (MRI).

MATERIALS AND METHODS

A total of 138 patients (age, 43.7 ± 10.3 years) with uterine sarcoma (n = 44) and atypical leiomyoma (n = 94) were retrospectively collected from four institutions. The cohort was randomly divided into training (84/138, 60.0%) and validation (54/138, 40.0%) sets. Two independent readers evaluated six qualitative MRI features and two DWI-based quantitative parameters for each index tumor. Multivariable logistic regression was used to identify the relevant qualitative MRI features. Diagnostic classifiers based on qualitative MRI features alone and in combination with DWI-based quantitative parameters were developed using a logistic regression algorithm. The diagnostic performance of the classifiers was evaluated using a cross-table analysis and calculation of the area under the receiver operating characteristic curve (AUC).

RESULTS

Mean apparent diffusion coefficient value of uterine sarcoma was lower than that of atypical leiomyoma (mean ± standard deviation, 0.94 ± 0.30 10-3 mm²/s vs. 1.23 ± 0.25 10-3 mm²/s; P < 0.001), and the relative contrast ratio was higher in the uterine sarcoma (8.16 ± 2.94 vs. 4.19 ± 2.66; P < 0.001). Selected qualitative MRI features included ill-defined margin (adjusted odds ratio [aOR], 17.9; 95% confidence interval [CI], 1.41-503, P = 0.040), intratumoral hemorrhage (aOR, 27.3; 95% CI, 3.74-596, P = 0.006), and absence of T2 dark area (aOR, 83.5; 95% CI, 12.4-1916, P < 0.001). The classifier that combined qualitative MRI features and DWI-based quantitative parameters showed significantly better performance than without DWI-based parameters in the validation set (AUC, 0.92 vs. 0.78; P < 0.001).

CONCLUSION

The addition of DWI-based quantitative parameters to qualitative MRI features improved the diagnostic performance of the logistic regression classifier in differentiating uterine sarcomas from atypical leiomyomas on preoperative MRI.

Survivin-Sodium Iodide Symporter Reporter as a Non-Invasive Diagnostic Marker to Differentiate Uterine Leiomyosarcoma from Leiomyoma

Leiomyosarcoma (LMS) has been challenging to diagnose because of limitations in clinical and radiographic predictors, as well as the lack of reliable serum or urinary biomarkers. Most uterine masses consist of benign leiomyoma (LM). However, it is currently a significant challenge in gynecology practice to differentiate LMS from LM. This inability poses grave consequences for patients, leading to a high number of unnecessary hysterectomies, infertility, and other major morbidities and possible mortalities. This study aimed to evaluate the use of Survivin-Sodium iodide symporter (Ad-Sur-NIS) as a reporter gene biomarker to differentiate malignant LMS from benign LM by using an F18-NaBF4 PET/CT scan. The PET/CT scan images showed a significantly increased radiotracer uptake and a decreased radiotracer decay attributable to the higher abundance of Ad-Sur-NIS in the LMS tumors compared to LM (p < 0.05). An excellent safety profile was observed, with no pathological or metabolic differences detected in Ad-Sur-NIS-treated animal versus the vehicle control. Ad-Sur-NIS as a PET scan reporter is a promising imaging biomarker that can differentiate uterine LMS from LM using F18-NaBF4 as a radiotracer. As a new diagnostic method, the F18 NaBF4 PET/CT scan can provide a much-needed tool in clinical practices to effectively triage women with suspicious uterine masses and avoid unnecessary invasive interventions.

A human-interpretable machine learning pipeline based on ultrasound to support leiomyosarcoma diagnosis

The preoperative evaluation of myometrial tumors is essential to avoid delayed treatment and to establish the appropriate surgical approach. Specifically, the differential diagnosis of leiomyosarcoma (LMS) is particularly challenging due to the overlapping of clinical, laboratory and ultrasound features between fibroids and LMS. In this work, we present a human-interpretable machine learning (ML) pipeline to support the preoperative differential diagnosis of LMS from leiomyomas, based on both clinical data and gynecological ultrasound assessment of 68 patients (8 with LMS diagnosis). The pipeline provides the following novel contributions: (i) end-users have been involved both in the definition of the ML tasks and in the evaluation of the overall approach; (ii) clinical specialists get a full understanding of both the decision-making mechanisms of the ML algorithms and the impact of the features on each automatic decision. Moreover, the proposed pipeline addresses some of the problems concerning both the imbalance of the two classes by analyzing and selecting the best combination of the synthetic oversampling strategy of the minority class and the classification algorithm among different choices, and the explainability of the features at global and local levels. The results show very high performance of the best strategy (AUC = 0.99, F1 = 0.87) and the strong and stable impact of two ultrasound-based features (i.e., tumor borders and consistency of the lesions). Furthermore, the SHAP algorithm was exploited to quantify the impact of the features at the local level and a specific module was developed to provide a template-based natural language (NL) translation of the explanations for enhancing their interpretability and fostering the use of ML in the clinical setting.

A contrast-enhanced CT-based radiomic nomogram for the differential diagnosis of intravenous leiomyomatosis and uterine leiomyoma

Objective

Uterine intravenous leiomyomatosis (IVL) is a rare and unique leiomyoma that is difficult to surgery due to its ability to extend into intra- and extra-uterine vasculature. And it is difficult to differentiate from uterine leiomyoma (LM) by conventional CT scanning, which results in a large number of missed diagnoses. This study aimed to evaluate the utility of a contrast-enhanced CT-based radiomic nomogram for preoperative differentiation of IVL and LM.

Methods

124 patients (37 IVL and 87 LM) were retrospectively enrolled in the study. Radiomic features were extracted from contrast-enhanced CT before surgery. Clinical, radiomic, and combined models were developed using LightGBM (Light Gradient Boosting Machine) algorithm to differentiate IVL and LM. The clinical and radiomic signatures were integrated into a nomogram. The diagnostic performance of the models was evaluated using the area under the curve (AUC) and decision curve analysis (DCA).

Results

Clinical factors, such as symptoms, menopausal status, age, and selected imaging features, were found to have significant correlations with the differential diagnosis of IVL and LM. A total of 108 radiomic features were extracted from contrast-enhanced CT images and selected for analysis. 29 radiomics features were selected to establish the Rad-score. A clinical model was developed to discriminate IVL and LM (AUC=0.826). Radiomic models were used to effectively differentiate IVL and LM (AUC=0.980). This radiological nomogram combined the Rad-score with independent clinical factors showed better differentiation efficiency than the clinical model (AUC=0.985, p=0.046).

Conclusion

This study provides evidence for the utility of a radiomic nomogram integrating clinical and radiomic signatures for differentiating IVL and LM with improved diagnostic accuracy. The nomogram may be useful in clinical decision-making and provide recommendations for clinical treatment.

Multiparametric MRI radiomics in prostate cancer for predicting Ki-67 expression and Gleason score: a multicenter retrospective study

PURPOSE

Prostate cancer (PCa) with high Ki-67 expression and high Gleason Scores (GS) tends to have aggressive clinicopathological characteristics and a dismal prognosis. In order to predict the Ki-67 expression status and the GS in PCa, we sought to construct and verify MRI-based radiomics signatures.

METHODS AND MATERIALS

We collected T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC) images from 170 PCa patients at three institutions and extracted 321 original radiomic features from each image modality. We used support vector machine (SVM) and least absolute shrinkage and selection operator (LASSO) logistic regression to select the most informative radiomic features and built predictive models using up sampling and feature selection techniques. Using receiver operating characteristic (ROC) analysis, the discriminating power of this feature was determined. Subsequent decision curve analysis (DCA) assessed the clinical utility of the radiomic features. The Kaplan-Meier (KM) test revealed that the radiomics-predicted Ki-67 expression status and GS were prognostic factors for PCa survival.

RESULT

The hypothesized radiomics signature, which included 15 and 9 selected radiomics features, respectively, was significantly correlated with pathological Ki-67 and GS outcomes in both the training and validation datasets. Areas under the curve (AUC) for the developed model were 0.813 (95% CI 0.681,0.930) and 0.793 (95% CI 0.621, 0.929) for the training and validation datasets, respectively, demonstrating discrimination and calibration performance. The model's clinical usefulness was verified using DCA. In both the training and validation sets, high Ki-67 expression and high GS predicted by radiomics using SVM models were substantially linked with poor overall survival (OS).

CONCLUSIONS

Both Ki-67 expression status and high GS correlate with PCa patient survival outcomes; therefore, the ability of the SVM classifier-based model to estimate Ki-67 expression status and the Lasso classifier-based model to assess high GS may enhance clinical decision-making.

Methodological considerations on segmenting rhabdomyosarcoma with diffusion-weighted imaging-What can we do better?

PURPOSE

Diffusion-weighted MRI is a promising technique to monitor response to treatment in pediatric rhabdomyosarcoma. However, its validation in clinical practice remains challenging. This study aims to investigate how the tumor segmentation strategy can affect the apparent diffusion coefficient (ADC) measured in pediatric rhabdomyosarcoma.

MATERIALS AND METHODS

A literature review was performed in PubMed using search terms relating to MRI and sarcomas to identify commonly applied segmentation strategies. Seventy-six articles were included, and their presented segmentation methods were evaluated. Commonly reported segmentation strategies were then evaluated on diffusion-weighted imaging of five pediatric rhabdomyosarcoma patients to assess their impact on ADC.

RESULTS

We found that studies applied different segmentation strategies to define the shape of the region of interest (ROI)(outline 60%, circular ROI 27%), to define the segmentation volume (2D 44%, multislice 9%, 3D 21%), and to define the segmentation area (excludes edge 7%, excludes other region 19%, specific area 27%, whole tumor 48%). In addition, details of the segmentation strategy are often unreported. When implementing and comparing these strategies on in-house data, we found that excluding necrotic, cystic, and hemorrhagic areas from segmentations resulted in on average 5.6% lower mean ADC. Additionally, the slice location used in 2D segmentation methods could affect ADC by as much as 66%.

CONCLUSION

Diffusion-weighted MRI studies in pediatric sarcoma currently employ a variety of segmentation methods. Our study shows that different segmentation strategies can result in vastly different ADC measurements, highlighting the importance to further investigate and standardize segmentation.

Joint EANM/SNMMI guideline on radiomics in nuclear medicine : Jointly supported by the EANM Physics Committee and the SNMMI Physics, Instrumentation and Data Sciences Council

PURPOSE

The purpose of this guideline is to provide comprehensive information on best practices for robust radiomics analyses for both hand-crafted and deep learning-based approaches.

METHODS

In a cooperative effort between the EANM and SNMMI, we agreed upon current best practices and recommendations for relevant aspects of radiomics analyses, including study design, quality assurance, data collection, impact of acquisition and reconstruction, detection and segmentation, feature standardization and implementation, as well as appropriate modelling schemes, model evaluation, and interpretation. We also offer an outlook for future perspectives.

CONCLUSION

Radiomics is a very quickly evolving field of research. The present guideline focused on established findings as well as recommendations based on the state of the art. Though this guideline recognizes both hand-crafted and deep learning-based radiomics approaches, it primarily focuses on the former as this field is more mature. This guideline will be updated once more studies and results have contributed to improved consensus regarding the application of deep learning methods for radiomics. Although methodological recommendations in the present document are valid for most medical image modalities, we focus here on nuclear medicine, and specific recommendations when necessary are made for PET/CT, PET/MR, and quantitative SPECT.

Combining multiparametric MRI features-based transfer learning and clinical parameters: application of machine learning for the differentiation of uterine sarcomas from atypical leiomyomas

OBJECTIVES

To explore the feasibility and effectiveness of machine learning (ML) based on multiparametric magnetic resonance imaging (mp-MRI) features extracted from transfer learning combined with clinical parameters to differentiate uterine sarcomas from atypical leiomyomas (ALMs).

METHODS

The data of 86 uterine sarcomas between July 2011 and December 2019 and 86 ALMs between June 2013 and June 2017 were retrospectively reviewed. We extracted deep-learning features and radiomics features from T2-weighted imaging (T2WI) and diffusion weighted imaging (DWI). The two feature extraction methods, transfer learning and radiomics, were compared. Random forest was adopted as the classifier. T2WI features, DWI features, combined T2WI and DWI (mp-MRI) features, and combined clinical parameters and mp-MRI features were applied to establish T2, DWI, T2-DWI, and complex multiparameter (mp) models, respectively. Predictive performance was assessed with the area under the receiver operating characteristic curve (AUC).

RESULTS

In the test set, the T2, DWI, T2-DWI and complex mp models based on transfer learning (AUCs range from 0.76 to 0.81, 0.80 to 0.88, 0.85 to 0.92, and 0.94 to 0.96, respectively) outperformed the models based on radiomics (AUCs of 0.73, 0.76, 0.79, and 0.92, respectively). Moreover, the complex mp model showed the best prediction performance, with the Resnet50-complex mp model achieving the highest AUC (0.96) and accuracy (0.87).

CONCLUSIONS

Transfer learning is feasible and superior to radiomics in the differential diagnosis of uterine sarcomas and ALMs in our dataset. ML models based on deep learning features of nonenhanced mp-MRI and clinical parameters can achieve good diagnostic efficacy.

KEY POINTS

• The ML model combining nonenhanced mp-MRI features and clinical parameters can distinguish uterine sarcomas from ALMs. • Transfer learning can be applied to differentiate uterine sarcomas from ALMs and outperform radiomics. • The most accurate prediction model was Resnet50-based transfer learning, built with the deep-learning features of mp-MRI and clinical parameters.

Preoperative Prediction of Inferior Vena Cava Wall Invasion of Tumor Thrombus in Renal Cell Carcinoma: Radiomics Models Based on Magnetic Resonance Imaging

Objective

To develop radiomics models to predict inferior vena cava (IVC) wall invasion by tumor thrombus (TT) in patients with renal cell carcinoma (RCC).

Methods

Preoperative MR images were retrospectively collected from 91 patients with RCC who underwent radical nephrectomy (RN) and thrombectomy. The images were randomly allocated into a training (n = 64) and validation (n = 27) cohort. The inter-and intra-rater agreements were organized to compare masks delineated by two radiologists. The masks of TT and IVC were manually annotated on axial fat-suppression T2-weighted images (fsT2WI) by one radiologist. The following models were trained to predict the probability of IVC wall invasion: two radiomics models using radiomics features extracted from the two masks (model 1, radiomics model_IVC; model 2, radiomics model_TT), two combined models using radiomics features and radiological features (model 3, combined model_IVC; model 4, combined model_TT), and one radiological model (model 5) using radiological features. Receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA) were applied to validate the discriminatory effect and clinical benefit of the models.

Results

Model 1 to model 5 yielded area under the curves (AUCs) of 0.881, 0.857, 0.883, 0.889, and 0.769, respectively, in the validation cohort. No significant differences were found between these models (p = 0.108-0.951). The dicision curve analysis (DCA) showed that the model 3 had a higher overall net benefit than the model 1, model 2, model 4, and model 5.

Conclusions

The combined model_IVC (model 3) based on axial fsT2WI exhibited excellent predictive performance in predicting IVC wall invasion status.

Preoperative Differentiation of Uterine Leiomyomas and Leiomyosarcomas: Current Possibilities and Future Directions

The distinguishing of uterine leiomyosarcomas (ULMS) and uterine leiomyomas (ULM) before the operation and histopathological evaluation of tissue is one of the current challenges for clinicians and researchers. Recently, a few new and innovative methods have been developed. However, researchers are trying to create different scales analyzing available parameters and to combine them with imaging methods with the aim of ULMs and ULM preoperative differentiation ULMs and ULM. Moreover, it has been observed that the technology, meaning machine learning models and artificial intelligence (AI), is entering the world of medicine, including gynecology. Therefore, we can predict the diagnosis not only through symptoms, laboratory tests or imaging methods, but also, we can base it on AI. What is the best option to differentiate ULM and ULMS preoperatively? In our review, we focus on the possible methods to diagnose uterine lesions effectively, including clinical signs and symptoms, laboratory tests, imaging methods, molecular aspects, available scales, and AI. In addition, considering costs and availability, we list the most promising methods to be implemented and investigated on a larger scale.

Radiomics and Artificial Intelligence in Uterine Sarcomas: A Systematic Review

Background: Recently, artificial intelligence (AI) with computerized imaging analysis is attracting the attention of clinicians, in particular for its potential applications in improving cancer diagnosis. This review aims to investigate the contribution of radiomics and AI on the radiological preoperative assessment of patients with uterine sarcomas (USs). Methods: Our literature review involved a systematic search conducted in the last ten years about diagnosis, staging and treatments with radiomics and AI in USs. The protocol was drafted according to the systematic review and meta-analysis preferred reporting project (PRISMA-P) and was registered in the PROSPERO database (CRD42021253535). Results: The initial search identified 754 articles; of these, six papers responded to the characteristics required for the revision and were included in the final analysis. The predominant technique tested was magnetic resonance imaging. The analyzed studies revealed that even though sometimes complex models included AI-related algorithms, they are still too complex for translation into clinical practice. Furthermore, since these results are extracted by retrospective series and do not include external validations, currently it is hard to predict the chances of their application in different study groups. Conclusion: To date, insufficient evidence supports the benefit of radiomics in USs. Nevertheless, this field is promising but the quality of studies should be a priority in these new technologies.

Accurate Tumor Delineation vs. Rough Volume of Interest Analysis for 18F-FDG PET/CT Radiomics-Based Prognostic Modeling inNon-Small Cell Lung Cancer

Background

The aim of this work was to investigate the ability of building prognostic models in non-small cell lung cancer (NSCLC) using radiomic features from positron emission tomography and computed tomography with 2-deoxy-2-[fluorine-18]fluoro-d-glucose (¹⁸F-FDG PET/CT) images based on a "rough" volume of interest (VOI) containing the tumor instead of its accurate delineation, which is a significant time-consuming bottleneck of radiomics analyses.

Methods

A cohort of 138 patients with stage II-III NSCLC treated with radiochemotherapy recruited retrospectively (n = 87) and prospectively (n = 51) was used. Two approaches were compared: firstly, the radiomic features were extracted from the delineated primary tumor volumes in both PET (using the automated fuzzy locally adaptive Bayesian, FLAB) and CT (using a semi-automated approach with 3D Slicer™) components. Both delineations were carried out within previously manually defined "rough" VOIs containing the tumor and the surrounding tissues, which were exploited for the second approach: the same features were extracted from this alternative VOI. Both sets for features were then combined with the clinical variables and processed through the same machine learning (ML) pipelines using the retrospectively recruited patients as the training set and the prospectively recruited patients as the testing set. Logistic regression (LR), random forest (RF), and support vector machine (SVM), as well as their consensus through averaging the output probabilities, were considered for feature selection and modeling for overall survival (OS) prediction as a binary classification (either median OS or 6 months OS). The resulting models were compared in terms of balanced accuracy, sensitivity, and specificity.

Results

Overall, better performance was achieved using the features from delineated tumor volumes. This was observed consistently across ML algorithms and for the two clinical endpoints. However, the loss of performance was not significant, especially when a consensus of the three ML algorithms was considered (0.89 vs. 0.88 and 0.78 vs. 0.77).

Conclusion

Our findings suggest that it is feasible to achieve similar levels of prognostic accuracy in radiomics-based modeling by relying on a faster and easier VOI definition, skipping a time-consuming tumor delineation step, thus facilitating automation of the whole radiomics workflow. The associated cost is a loss of performance in the resulting models, although this loss can be greatly mitigated when a consensus of several models is relied upon.

New imaging modalities to distinguish rare uterine mesenchymal cancers from benign uterine lesions

PURPOSE OF REVIEW

Uterine sarcomas are rare and are often challenging to differentiate on imaging from benign mimics, such as leiomyoma. As functional MRI techniques have improved and new adjuncts, such as machine learning and texture analysis, are now being investigated, it is helpful to be aware of the current literature on imaging features that may sometimes allow for preoperative distinction.

RECENT FINDINGS

MRI, with both conventional and functional imaging, is the modality of choice for evaluating uterine mesenchymal tumors, especially in differentiating uterine leiomyosarcoma from leiomyoma through validated diagnostic algorithms. MRI is sometimes helpful in differentiating high-grade stromal sarcoma from low-grade stromal sarcoma or differentiating endometrial stromal sarcoma from endometrial carcinoma. However, imaging remains nonspecific for evaluating rarer neoplasms, such as uterine tumor resembling ovarian sex cord tumor or perivascular epithelioid cell tumor, primarily because of the small number and power of relevant studies.

SUMMARY

Through advances in MRI techniques and novel investigational imaging adjuncts, such as machine learning and texture analysis, imaging differentiation of malignant from benign uterine mesenchymal tumors has improved and could help reduce morbidity relating to misdiagnosis or diagnostic delays.

Radiomics in the Setting of Neoadjuvant Radiotherapy: A New Approach for Tailored Treatment

INTRODUCTION

Neoadjuvant radiotherapy is currently used mainly in locally advanced rectal cancer and sarcoma and in a subset of non-small cell lung cancer and esophageal cancer, whereas in other diseases it is under investigation. The evaluation of the efficacy of the induction strategy is made possible by performing imaging investigations before and after the neoadjuvant therapy and is usually challenging. In the last decade, texture analysis (TA) has been developed to help the radiologist to quantify and identify the parameters related to tumor heterogeneity, which cannot be appreciated by the naked eye. The aim of this narrative is to review the impact of TA on the prediction of response to neoadjuvant radiotherapy and or chemoradiotherapy.

MATERIALS AND METHODS

Key references were derived from a PubMed query. Hand searching and ClinicalTrials.gov were also used.

RESULTS

This paper contains a narrative report and a critical discussion of radiomics approaches in different fields of neoadjuvant radiotherapy, including esophageal cancer, lung cancer, sarcoma, and rectal cancer.

CONCLUSIONS

Radiomics can shed a light on the setting of neoadjuvant therapies that can be used to tailor subsequent approaches or even to avoid surgery in the future. At the same, these results need to be validated in prospective and multicenter trials.

Preoperative Prediction of Extracapsular Extension: Radiomics Signature Based on Magnetic Resonance Imaging to Stage Prostate Cancer

PURPOSE

To investigate and validate the potential role of a radiomics signature in predicting the side-specific probability of extracapsular extension (ECE) of prostate cancer (PCa).

PROCEDURES

The preoperative magnetic resonance imaging data of 238 prostatic samples from 119 enrolled PCa patients were retrospectively assessed. The samples with were randomized in a two-to-one ratio into training (n = 74) and validation (n = 45) datasets. The radiomics features were derived from T2-weighted images (T2WIs). The optimal radiomics features were identified from the least absolute shrinkage and selection operator (LASSO) logistic regression model and were used to construct a predictive radiomics signature via dimension reduction and selection approaches. The association between the radiomics signatures and pathological ECE status was explored. Receiver operating characteristic (ROC) analysis was used to assess the discriminatory ability of the signature. The calibration performance and clinical usefulness of the radiomics signature were subsequently assessed by calibration curve and decision curve analyses.

RESULTS

The proposed radiomics signature that incorporated 17 selected radiomics features was significantly associated with pathological ECE outcomes (P < 0.001) in both the training and validation datasets. The constructed model displayed good discrimination, with areas under the curve (AUC) of 0.906 (95 % confidence interval (CI), 0.847, 0.948) and 0.821 (95 % CI, 0.726, 0.894) for the training and validation datasets, respectively, and had a good calibration performance. The clinical utility of this model was confirmed through decision curve analysis.

CONCLUSIONS

The radiomics signature based on T2WIs showed the potential to predict the side-specific probability of pathological ECE status and can facilitate the preoperative individualized predictions for PCa patients.

Whole-tumor 3D volumetric MRI-based radiomics approach for distinguishing between benign and malignant soft tissue tumors

OBJECTIVES

Our purpose was to differentiate between malignant from benign soft tissue neoplasms using a combination of MRI-based radiomics metrics and machine learning.

METHODS

Our retrospective study identified 128 histologically diagnosed benign (n = 36) and malignant (n = 92) soft tissue lesions. 3D ROIs were manually drawn on 1 sequence of interest and co-registered to other sequences obtained during the same study. One thousand seven hundred eight radiomics features were extracted from each ROI. Univariate analyses with supportive ROC analyses were conducted to evaluate the discriminative power of predictive models constructed using Real Adaptive Boosting (Adaboost) and Random Forest (RF) machine learning approaches.

RESULTS

Univariate analyses demonstrated that 36.89% of individual radiomics varied significantly between benign and malignant lesions at the p ≤ 0.05 level. Adaboost and RF performed similarly well, with AUCs of 0.77 (95% CI 0.68-0.85) and 0.72 (95% CI 0.63-0.81), respectively, after 10-fold cross-validation. Restricting the machine learning models to only sequences extracted from T2FS and STIR sequences maintained comparable performance, with AUCs of 0.73 (95% CI 0.64-0.82) and 0.75 (95% CI 0.65-0.84), respectively.

CONCLUSION

Machine learning decision classifiers constructed from MRI-based radiomics features show promising ability to preoperatively discriminate between benign and malignant soft tissue masses. Our approach maintains applicability even when the dataset is restricted to T2FS and STIR fluid-sensitive sequences, which may bolster practicality in clinical application scenarios by eliminating the need for complex co-registrations for multisequence analysis.

KEY POINTS

• Predictive models constructed from MRI-based radiomics data and machine learning-augmented approaches yielded good discriminative power to correctly classify benign and malignant lesions on preoperative scans, with AUCs of 0.77 (95% CI 0.68-0.85) and 0.72 (95% CI 0.63-0.81) for Real Adaptive Boosting (Adaboost) and Random Forest (RF), respectively. • Restricting the models to only use metrics extracted from T2 fat-saturated (T2FS) and Short-Tau Inversion Recovery (STIR) sequences yielded similar performance, with AUCs of 0.73 (95% CI 0.64-0.82) and 0.75 (95% CI 0.65-0.84) for Adaboost and RF, respectively. • Radiomics-based machine learning decision classifiers constructed from multicentric data more closely mimic the real-world practice environment and warrant additional validation ahead of prospective implementation into clinical workflows.

Unmet Medical Needs and Future Perspectives for Leiomyosarcoma Patients-A Position Paper from the National LeioMyoSarcoma Foundation (NLMSF) and Sarcoma Patients EuroNet (SPAEN)

As leiomyosarcoma patients are challenged by the development of metastatic disease, effective systemic therapies are the cornerstone of outcome. However, the overall activity of the currently available conventional systemic treatments and the prognosis of patients with advanced or metastatic disease are still poor, making the treatment of this patient group challenging. Therefore, in a joint effort together with patient networks and organizations, namely Sarcoma Patients EuroNet (SPAEN), the international network of sarcoma patients organizations, and the National LeioMyoSarcoma Foundation (NLMSF) in the United States, we aim to summarize state-of-the-art treatments for leiomyosarcoma patients in order to identify knowledge gaps and current unmet needs, thereby guiding the community to design innovative clinical trials and basic research and close these research gaps. This position paper arose from a leiomyosarcoma research meeting in October 2020 hosted by the NLMSF and SPAEN.

Preoperative differentiation of pancreatic mucinous cystic neoplasm from macrocystic serous cystic adenoma using radiomics: Preliminary findings and comparison with radiological model

PURPOSE

To develop a radiomics model in the preoperative differentiation of mucinous cystic neoplasm (MCN) and macrocystic serous cystadenoma (MaSCA) and to compare its diagnostic performance with conventional radiological model.

METHODS

57 Patients (MCN = 31, MaSCA = 26) with preoperative multidetector computed tomography (MDCT) scans were retrospectively included in this study. A radiological model was constructed from radiological features evaluated by radiologists. A radiomics model was constructed with high-dimensional quantitative features extracted from manually segmented volume of interests (VOIs). A combined model was constructed using both radiomics features and radiological features. The diagnostic performance of three models were assessed by the area under the receiver-operating characteristic curve (AUC), sensitivity, specificity, accuracy, and the calibration curves.

RESULTS

The radiological model yielded an AUC of 0.775, sensitivity of 74.2 %, specificity of 80.8, and accuracy of 77.2 %. The radiomics model yielded an AUC of 0.989, sensitivity of 93.6 %, specificity of 96.2 %, and accuracy of 94.7 %. The combined model yielded an AUC of 0.994, sensitivity of 96.8 %, specificity of 100 %, and accuracy of 98.2 %. Both combined model and radiomics model showed higher AUC, sensitivity, and accuracy than radiological model (all P <  .05). The combined model showed higher AUC than radiomics model, though no significant difference was found (P =  .41). The combined model showed better calibration than radiomics model (P =  .91 vs. P <  .001).

CONCLUSIONS

Combined model which contained both radiomics features and radiological features outperformed radiomics model and radiological model in the preoperative differentiation of MCN and MaSCA.

MRI-Based Radiomics Signature for the Preoperative Prediction of Extracapsular Extension of Prostate Cancer

BACKGROUND

Radiomics approaches based on multiparametric MRI (mp-MRI) have shown high accuracy in prostate cancer (PCa) management. However, there is a need to apply radiomics to the preoperative prediction of extracapsular extension (ECE).

PURPOSE

To develop and validate a radiomics signature to preoperatively predict the probability of ECE for patients with PCa, compared with the radiologists' interpretations.

STUDY TYPE

Retrospective.

POPULATION

In total, 210 patients with pathology-confirmed ECE status (101 positive, 109 negative) were enrolled.

FIELD STRENGTH/SEQUENCE

T₂ -weighted imaging (T₂ WI), diffusion-weighted imaging, and dynamic contrast-enhanced imaging were performed on two 3.0T MR scanners.

ASSESSMENT

A radiomics signature was constructed to predict the probability of ECE prior to radical prostatectomy (RP). In all, 17 stable radiomics features of 1619 extracted features based on T₂ WI were selected. The same images were also evaluated by three radiologists. The predictive performance of the radiomics signature was validated and compared with radiologists' interpretations.

STATISTICAL TESTS

A radiomics signature was developed by a least absolute shrinkage and selection operator (LASSO) regression algorithm. Samples enrolled were randomly divided into two groups (143 for training and 67 for validation). Discrimination, calibration, and clinical usefulness were validated by analysis of the receiver operating characteristic (ROC) curve, calibration curve, and the decision curve, respectively. The predictive performance was then compared with visual assessments of three radiologists.

RESULTS

The radiomics signature yielded an AUC of 0.902 and 0.883 in the training and validation cohort, respectively, and outperformed the visual assessment (AUC: 0.600-0.697) in the validation cohort. Pairwise comparisons demonstrated that the radiomics signature was more sensitive than the radiologists (75.00% vs. 46.88%-50.00%, all P < 0.05), but obtained comparable specificities (91.43% vs. (88.57%-94.29%); P ranged from 0.64-1.00).

DATA CONCLUSION

A radiomics signature was developed and validated that outperformed the radiologists' visual assessments in predicting ECE status.

LEVEL OF EVIDENCE

4 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2019;50:1914-1925.