Multiparametric MRI Imaging Research Articles

Comparison of data fusion strategies for automated prostate lesion detection using mpMRI correlated with whole mount histology

BackgroundIn this work, we compare input level, feature level and decision level data fusion techniques for automatic detection of clinically significant prostate lesions (csPCa).MethodsMultiple deep learning CNN architectures were developed using the Unet as the baseline. The CNNs use both multiparametric MRI images (T2W, ADC, and High b-value) and quantitative clinical data (prostate specific antigen (PSA), PSA density (PSAD), prostate gland volume & gross tumor volume (GTV)), and only mp-MRI images (n = 118), as input. In addition, co-registered ground truth data from whole mount histopathology images (n = 22) were used as a test set for evaluation.ResultsThe CNNs achieved for early/intermediate / late level fusion a precision of 0.41/0.51/0.61, recall value of 0.18/0.22/0.25, an average precision of 0.13 / 0.19 / 0.27, and F scores of 0.55/0.67/ 0.76. Dice Sorensen Coefficient (DSC) was used to evaluate the influence of combining mpMRI with parametric clinical data for the detection of csPCa. We compared the DSC between the predictions of CNN’s trained with mpMRI and parametric clinical and the CNN’s trained with only mpMRI images as input with the ground truth. We obtained a DSC of data 0.30/0.34/0.36 and 0.26/0.33/0.34 respectively. Additionally, we evaluated the influence of each mpMRI input channel for the task of csPCa detection and obtained a DSC of 0.14 / 0.25 / 0.28.ConclusionThe results show that the decision level fusion network performs better for the task of prostate lesion detection. Combining mpMRI data with quantitative clinical data does not show significant differences between these networks (p = 0.26/0.62/0.85). The results show that CNNs trained with all mpMRI data outperform CNNs with less input channels which is consistent with current clinical protocols where the same input is used for PI-RADS lesion scoring.Trial registrationThe trial was registered retrospectively at the German Register for Clinical Studies (DRKS) under proposal number Nr. 476/14 & 476/19.

Computer-aided detection of prostate cancer in early stages using multi-parameter MRI: A promising approach for early diagnosis.

Transrectal ultrasound-guided prostate biopsy is the gold standard diagnostic test for prostate cancer, but it is an invasive examination of non-targeted puncture and has a high false-negative rate. In this study, we aimed to develop a computer-assisted prostate cancer diagnosis method based on multiparametric MRI (mpMRI) images. We retrospectively collected 106 patients who underwent radical prostatectomy after diagnosis with prostate biopsy. mpMRI images, including T2 weighted imaging (T2WI), diffusion weighted imaging (DWI), and dynamic-contrast enhanced (DCE), and were accordingly analyzed. We extracted the region of interest (ROI) about the tumor and benign area on the three sequential MRI axial images at the same level. The ROI data of 433 mpMRI images were obtained, of which 202 were benign and 231 were malignant. Of those, 50 benign and 50 malignant images were used for training, and the 333 images were used for verification. Five main feature groups, including histogram, GLCM, GLGCM, wavelet-based multi-fractional Brownian motion features and Minkowski function features, were extracted from the mpMRI images. The selected characteristic parameters were analyzed by MATLAB software, and three analysis methods with higher accuracy were selected. Through prostate cancer identification based on mpMRI images, we found that the system uses 58 texture features and 3 classification algorithms, including Support Vector Machine (SVM), K-nearest Neighbor (KNN), and Ensemble Learning (EL), performed well. In the T2WI-based classification results, the SVM achieved the optimal accuracy and AUC values of 64.3% and 0.67. In the DCE-based classification results, the SVM achieved the optimal accuracy and AUC values of 72.2% and 0.77. In the DWI-based classification results, the ensemble learning achieved optimal accuracy as well as AUC values of 75.1% and 0.82. In the classification results based on all data combinations, the SVM achieved the optimal accuracy and AUC values of 66.4% and 0.73. The proposed computer-aided diagnosis system provides a good assessment of the diagnosis of the prostate cancer, which may reduce the burden of radiologists and improve the early diagnosis of prostate cancer.

Multiparametric Characterization of the DSL-6A/C1 Pancreatic Cancer Model in Rats.

The DSL-6A/C1 murine pancreatic ductal adenocarcinoma (PDAC) tumor model was established in Lewis rats and characterized through a comprehensive multiparametric analysis to compare it to other preclinical tumor models and explore potential diagnostic and therapeutical targets. DSL-6A/C1 tumors were histologically analyzed to elucidate PDAC features. The tumor microenvironment was studied for immune cell prevalence. Multiparametric MRI and PET imaging were utilized to characterize tumors, and 68Ga-FAPI-46-targeting cancer-associated fibroblasts (CAFs), were used to validate the histological findings. The histology confirmed typical PDAC characteristics, such as malformed pancreatic ductal malignant cells and CAFs. Distinct immune landscapes were identified, revealing an increased presence of CD8+ T cells and a decreased CD4+ T cell fraction within the tumor microenvironment. PET imaging with 68Ga-FAPI tracers exhibited strong tracer uptake in tumor tissues. The MRI parameters indicated increasing intralesional necrosis over time and elevated contrast media uptake in vital tumor areas. We have demonstrated that the DSL-6A/C1 tumor model, particularly due to its high tumorigenicity, tumor size, and 68Ga-FAPI-46 sensitivity, is a suitable alternative to established small animal models for many forms of preclinical analyses and therapeutic studies of PDAC.

Determination of p53abn endometrial cancer: a multitask analysis using radiological-clinical nomogram on MRI.

We aimed to differentiate endometrial cancer (EC) between TP53mutation (P53abn) and Non-P53abn subtypes using radiological-clinical nomogram on EC body volume MRI. We retrospectively recruited 227 patients with pathologically proven EC from our institution. All these patients have undergone molecular pathology diagnosis based on the Cancer Genome Atlas. Clinical characteristics and histological diagnosis were recorded from the hospital information system. Radiomics features were extracted from online Pyradiomics processors. The diagnostic performance across different acquisition protocols was calculated and compared. The radiological-clinical nomogram was established to determine the nonendometrioid, high-risk, and P53abn EC group. The best MRI sequence for differentiation P53abn from the non-P53abn group was contrast-enhanced T1WI (test AUC: 0.8). The best MRI sequence both for differentiation endometrioid cancer from nonendometrioid cancer and high-risk from low- and intermediate-risk groups was apparent diffusion coefficient map (test AUC: 0.665 and 0.690). For all 3 tasks, the combined model incorporating all the best discriminative features from each sequence yielded the best performance. The combined model achieved an AUC of 0.845 in the testing cohorts for P53abn cancer identification. The MR-based radiomics diagnostic model performed better than the clinical-based model in determining P53abn EC (AUC: 0.834 vs 0.682). In the present study, the diagnostic model based on the combination of both radiomics and clinical features yielded a higher performance in differentiating nonendometrioid and P53abn cancer from other EC molecular subgroups, which might help design a tailed treatment, especially for patients with high-risk EC. (1) The contrast-enhanced T1WI was the best MRI sequence for differentiation P53abn from the non-P53abn group (test AUC: 0.8). (2) The radiomics-based diagnostic model performed better than the clinical-based model in determining P53abn EC (AUC: 0.834 vs 0.682). (3) The proposed model derived from multi-parametric MRI images achieved a higher accuracy in P53abn EC identification (AUC: 0.845).

Fully Automated Identification of Lymph Node Metastases and Lymphovascular Invasion in Endometrial Cancer From Multi-Parametric MRI by Deep Learning.

Early and accurate identification of lymphatic node metastasis (LNM) and lymphatic vascular space invasion (LVSI) for endometrial cancer (EC) patients is important for treatment design, but difficult on multi-parametric MRI (mpMRI) images. To develop a deep learning (DL) model to simultaneously identify of LNM and LVSI of EC from mpMRI images. Retrospective. Six hundred twenty-one patients with histologically proven EC from two institutions, including 111 LNM-positive and 168 LVSI-positive, divided into training, internal, and external test cohorts of 398, 169, and 54 patients, respectively. T2-weighted imaging (T2WI), contrast-enhanced T1WI (CE-T1WI), and diffusion-weighted imaging (DWI) were scanned with turbo spin-echo, gradient-echo, and two-dimensional echo-planar sequences, using either a 1.5 T or 3 T system. EC lesions were manually delineated on T2WI by two radiologists and used to train an nnU-Net model for automatic segmentation. A multi-task DL model was developed to simultaneously identify LNM and LVSI positive status using the segmented EC lesion regions and T2WI, CE-T1WI, and DWI images as inputs. The performance of the model for LNM-positive diagnosis was compared with those of three radiologists in the external test cohort. Dice similarity coefficient (DSC) was used to evaluate segmentation results. Receiver Operating Characteristic (ROC) analysis was used to assess the performance of LNM and LVSI status identification. P value <0.05 was considered significant. EC lesion segmentation model achieved mean DSC values of 0.700 ± 0.25 and 0.693 ± 0.21 in the internal and external test cohorts, respectively. For LNM positive/LVSI positive identification, the proposed model achieved AUC values of 0.895/0.848, 0.806/0.795, and 0.804/0.728 in the training, internal, and external test cohorts, respectively, and better than those of three radiologists (AUC = 0.770/0.648/0.674). The proposed model has potential to help clinicians to identify LNM and LVSI status of EC patients and improve treatment planning. 3 TECHNICAL EFFICACY: Stage 2.

Deep learning based joint fusion approach to exploit anatomical and functional brain information in autism spectrum disorders

Background:The integration of the information encoded in multiparametric MRI images can enhance the performance of machine-learning classifiers. In this study, we investigate whether the combination of structural and functional MRI might improve the performances of a deep learning (DL) model trained to discriminate subjects with Autism Spectrum Disorders (ASD) with respect to typically developing controls (TD).Material and methodsWe analyzed both structural and functional MRI brain scans publicly available within the ABIDE I and II data collections. We considered 1383 male subjects with age between 5 and 40 years, including 680 subjects with ASD and 703 TD from 35 different acquisition sites. We extracted morphometric and functional brain features from MRI scans with the Freesurfer and the CPAC analysis packages, respectively. Then, due to the multisite nature of the dataset, we implemented a data harmonization protocol. The ASD vs. TD classification was carried out with a multiple-input DL model, consisting in a neural network which generates a fixed-length feature representation of the data of each modality (FR-NN), and a Dense Neural Network for classification (C-NN). Specifically, we implemented a joint fusion approach to multiple source data integration. The main advantage of the latter is that the loss is propagated back to the FR-NN during the training, thus creating informative feature representations for each data modality. Then, a C-NN, with a number of layers and neurons per layer to be optimized during the model training, performs the ASD-TD discrimination. The performance was evaluated by computing the Area under the Receiver Operating Characteristic curve within a nested 10-fold cross-validation. The brain features that drive the DL classification were identified by the SHAP explainability framework.ResultsThe AUC values of 0.66±0.05 and of 0.76±0.04 were obtained in the ASD vs. TD discrimination when only structural or functional features are considered, respectively. The joint fusion approach led to an AUC of 0.78±0.04. The set of structural and functional connectivity features identified as the most important for the two-class discrimination supports the idea that brain changes tend to occur in individuals with ASD in regions belonging to the Default Mode Network and to the Social Brain.ConclusionsOur results demonstrate that the multimodal joint fusion approach outperforms the classification results obtained with data acquired by a single MRI modality as it efficiently exploits the complementarity of structural and functional brain information.

The Diagnostic Value of a Nomogram Based on Clinical Imaging and MRIBased Radiomic Features in Triple-Negative Breast Cancer.

This study aimed to determine the utility of a radiomic nomogram combined with clinical imaging and radiomic features based on MRI for the diagnosis of triple-negative breast cancer. Multi-parametric MRI images of 136 breast cancer patients were retrospectively analyzed, 95 cases were stratified into the training cohort, and 41 cases were selected for the test group. According to the pathological molecular typing, the patients were divided into 23 cases of triple-negative breast cancer and 113 cases of non-triple-negative breast cancer. ITK software was used to manually delineate the lesion volume region of interest (VOI), and the Pyradiomics package was used to extract radiomic features for screening and model building. The platform was then used to analyze the clinical and imaging risk factors of breast cancer to build a characteristic model separately. Finally, a radiomic nomogram was constructed by integrating the radiomic and independent clinical image features. The diagnostic performance of the model was assessed using ROC curves. Univariate and multivariate analyses showed that the menstrual cycle, glandular density, and skin thickening were risk factors for clinical imaging characteristics of triple-negative breast cancer. The Area Under the Curve (AUC) was 0.839 and 0.826 for univariate and multivariate analysis, respectively. After screening, 11 radiomic features participated in the calculation of the radiomic score, and its AUC in the test set was 0.803. Combining it further with clinical models, the AUC improved to 0.899. The radiomic nomogram developed in this study has great value in the diagnosis of triple-negative breast cancer.

Multiparametric MRI and T2/FLAIR mismatch complements the World Health Organization 2021 classification for the diagnosis of IDH-mutant 1p/19q non-co-deleted/ATRX-mutant astrocytoma

To investigate whether T2-weighted imaging-fluid-attenuated inversion recovery (T2/FLAIR) mismatch, T2∗ dynamic susceptibility contrast (DSC) perfusion, and magnetic resonance spectroscopy (MRS) correlated with the histological diagnosis and grading of IDH (isocitrate dehydrogenase)-mutant, 1p/19q non-co-deleted/ATRX (alpha-thalassemia mental retardation X-linked)-mutant astrocytoma. Imaging of 101 IDH-mutant diffuse glioma cases of histological grades 2-3 (2019-2021) were analysed retrospectively by two neuroradiologists blinded to the molecular diagnosis. T2/FLAIR mismatch sign is used for radio-phenotyping, and pre-biopsy multiparametric MRI images were assessed for grading purposes. Cut-off values pre-determined for radiologically high-grade lesions were relative cerebral blood volume (rCBV) ≥2, choline/creatine ratio (Cho/Cr) ≥1.5 (30 ms echo time [TE]), Cho/Cr ≥1.8 (135 ms TE). Sixteen of the 101 cases showed T2/FLAIR mismatch, all of which were histogenetically confirmed IDH-mutant 1p/19q non-co-deleted/ATRX mutant astrocytomas; 50% were grade 3 (8/16) and 50% grade 2 (8/16). None showed contrast enhancement. Nine of the 16 had adequate multiparametric MRI for analysis. Any positive value by combining rCBV ≥2 with Cho/Cr ≥1.5 (30 ms TE) or Cho/Cr ≥1.8 (135 ms TE) predicted grade 3 histology with sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 100%. The T2/FLAIR mismatch sign detected diffuse astrocytomas with 100% specificity. When combined with high Cho/Cr and raised rCBV, this predicted histological grading with high accuracy. The future direction for imaging should explore a similar integrated layered approach of 2021 classification of central nervous system (CNS) tumours combining radio-phenotyping and grading from structural and multiparametric imaging.

Development and validation of a nomogram for predicting prostate cancer based on combining contrast-enhanced transrectal ultrasound and biparametric MRI imaging.

This study was to explore the feasibility of combining contrast-enhanced transrectal ultrasound (CE-TRUS) with biparametric MRI (CEUS-BpMRI) score for diagnosing prostate cancer (PCa). A total of 183 patients with suspected PCa who underwent multiparametric MRI (Mp-MRI) and CE-TRUS were included. CEUS-BpMRI score was developed based on the results of Mp-MRI and CE-TRUS. The diagnostic performance was evaluated by the area under the curve (AUC). The diagnostic efficacy of the CEUS-BpMRI score, BpMRI score, and PI-RADS v2.1 score were compared. Total patients were randomly assigned to a training cohort (70%) or validation cohort (30%). A nomogram was constructed based on univariate and multivariate logistic regression. The model was evaluated by AUC and calibration curve. The diagnostic performance of CEUS-BpMRI score (AUC 0.857) was comparable to that of PI-RADS v2.1 (AUC 0.862) (P = 0.499), and both were superior to Bp-MRI score (AUC 0.831, P < 0.05). In peripheral zone lesions with Bp-MRI score of 3, there was no statistically significant difference between PI-RADS v2.1 score (AUC 0.728) and CEUS-BpMRI score (AUC 0.668) (P = 0.479). Multivariate analysis showed that age, total prostate specific antigen/free prostate specific antigen (F/T), time to peak (TTP), and CEUS-BpMRI score were independent factors. The AUC of the nomogram was 0.909 in the training cohort and 0.914 in the validation cohort. CEUS-BpMRI score has high diagnostic efficacy for diagnosing PCa. A nomogram model established by combining age, F/T, TTP, and CEUS-BpMRI score can achieve the best predictive accuracy for PCa.

Multiparametric MRI radiomics for the differentiation of brain glial cell hyperplasia from low-grade glioma

BackgroundDifferentiating between low-grade glioma and brain glial cell hyperplasia is crucial for the customized clinical treatment of patients.ObjectiveBased on multiparametric MRI imaging and clinical risk factors, a radiomics-clinical model and nomogram were constructed for the distinction of brain glial cell hyperplasia from low-grade glioma.MethodsPatients with brain glial cell hyperplasia and low-grade glioma who underwent surgery at the First Affiliated Hospital of Soochow University from March 2016 to March 2022 were retrospectively included. In this study, A total of 41 patients of brain glial cell hyperplasia and 87 patients of low-grade glioma were divided into training group and validation group randomly at a ratio of 7:3. Radiomics features were extracted from T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), contrast-enhanced T1-weighted imaging (T1-enhanced). Then, LASSO, SVM, and RF models were created in order to choose a model with a greater level of efficiency for calculating each patient’s Rad-score (radiomics score). The independent risk factors were identified via univariate and multivariate logistic regression analysis to filter the Rad-score and clinical risk variables in turn. A radiomics-clinical model was next built of which effectiveness was assessed.ResultsBrain glial cell hyperplasia and low-grade gliomas from the 128 cases were randomly divided into 10 groups, of which 7 served as training group and 3 as validation group. The mass effect and Rad-score were two independent risk variables used in the construction of the radiomics-clinical model, and their respective AUCs for the training group and validation group were 0.847 and 0.858. The diagnostic accuracy, sensitivity, and specificity of the validation group were 0.821, 0.750, and 0.852 respectively.ConclusionCombining with radiomics constructed by multiparametric MRI images and clinical features, the radiomics-clinical model and nomogram that were developed to distinguish between brain glial cell hyperplasia and low-grade glioma had a good performance.

AI-based classification of three common malignant tumors in neuro-oncology: A multi-institutional comparison of machine learning and deep learning methods

PurposeTo determine if machine learning (ML) or deep learning (DL) pipelines perform better in AI-based three-class classification of glioblastoma (GBM), intracranial metastatic disease (IMD) and primary CNS lymphoma (PCNSL). MethodologyRetrospective analysis included 502 cases for training (208 GBM, 67 PCNSL and 227 IMD), with external validation on 86 cases (27:27:32). Multiparametric MRI images (T1W, T2W, FLAIR, DWI and T1-CE) were co-registered, resampled, denoised and intensity normalized, followed by semiautomatic 3D segmentation of the enhancing tumor (ET) and peritumoral region (PTR). Model performance was assessed using several ML pipelines and 3D-convolutional neural networks (3D-CNN) using sequence specific masks, as well as combination of masks. All pipelines were trained and evaluated with 5-fold nested cross-validation on internal data followed by external validation using multi-class AUC. ResultsTwo ML models achieved similar performance on test set, one using T2-ET and T2-PTR masks (AUC: 0.885, 95% CI: [0.816, 0.935] and another using T1-CE-ET and FLAIR-PTR mask (AUC: 0.878, CI: [0.804, 0.930]). The best performing DL models achieved an AUC of 0.854, (CI [0.774, 0.914]) on external data using T1-CE-ET and T2-PTR masks, followed by model derived from T1-CE-ET, ADC-ET and FLAIR-PTR masks (AUC: 0.851, CI [0.772, 0.909]). ConclusionBoth ML and DL derived pipelines achieved similar performance. T1-CE mask was used in three of the top four overall models. Additionally, all four models had some mask derived from PTR, either T2WI or FLAIR.

Machine-Learning-Based Radiomics for Classifying Glioma Grade from Magnetic Resonance Images of the Brain.

Grading of gliomas is a piece of critical information related to prognosis and survival. Classifying glioma grade by semantic radiological features is subjective, requires multiple MRI sequences, is quite complex and clinically demanding, and can very often result in erroneous radiological diagnosis. We used a radiomics approach with machine learning classifiers to determine the grade of gliomas. Eighty-three patients with histopathologically proven gliomas underwent MRI of the brain. Whenever available, immunohistochemistry was additionally used to augment the histopathological diagnosis. Segmentation was performed manually on the T2W MR sequence using the TexRad texture analysis softwareTM, Version 3.10. Forty-two radiomics features, which included first-order features and shape features, were derived and compared between high-grade and low-grade gliomas. Features were selected by recursive feature elimination using a random forest algorithm method. The classification performance of the models was measured using accuracy, precision, recall, f1 score, and area under the curve (AUC) of the receiver operating characteristic curve. A 10-fold cross-validation was adopted to separate the training and the test data. The selected features were used to build five classifier models: support vector machine, random forest, gradient boost, naive Bayes, and AdaBoost classifiers. The random forest model performed the best, achieving an AUC of 0.81, an accuracy of 0.83, f1 score of 0.88, a recall of 0.93, and a precision of 0.85 for the test cohort. The results suggest that machine-learning-based radiomics features extracted from multiparametric MRI images can provide a non-invasive method for predicting glioma grades preoperatively. In the present study, we extracted the radiomics features from a single cross-sectional image of the T2W MRI sequence and utilized these features to build a fairly robust model to classify low-grade gliomas from high-grade gliomas (grade 4 gliomas).

An Interactive App with Multi-parametric MRI – Whole-Mount Histology Correlation for Enhanced Prostate MRI Training of Radiology Residents

To validate the educational value of a newly created learning application in enhancing prostate MRI training of radiologists for detecting prostate cancer using an observer study. An interactive learning app, LearnRadiology, was developed using a web-based framework to display multi-parametric prostate MRI images with whole-mount histology for 20 cases curated for unique pathology and teaching points. Twenty new prostate MRI cases, different from the ones used in the web app, were uploaded on 3D Slicer. Three radiologists (R1: radiologist; R2, R3: residents) blinded to pathology results were asked to mark areas suspected of cancer and provide a confidence score (1-5, with 5 being high confidence level). Then after a minimum memory washout period of 1 month, the same radiologists used the learning app and then repeated the same observer study. The diagnostic performance for detecting cancers before and after accessing the learning app was measured by correlating MRI with whole-mount pathology by an independent reviewer. The 20 subjects included in the observer study had 39 cancer lesions (13 Gleason 3+3, 17 Gleason 3+4, 7 Gleason 4+3, and 2 Gleason 4+5 lesions). The sensitivity (R1: 54%→64%, P=0.08; R2: 44%→59%, P=0.03; R3: 62%→72%, P=0.04) and positive predictive value(R1: 68%→76%, P=0.23; R2: 52%→79%, P=0.01; R3: 48%→65%, P=0.04) for all 3 radiologists improved after using the teaching app. The confidence score for true positive cancer lesion also improved significantly (R1: 4.0±1.0→4.3±0.8; R2: 3.1±0.8→4.0±1.1; R3: 2.8±1.2→4.1±1.1; P<0.05). The web-based and interactive LearnRadiology app learning resource can support medical student and postgraduate education by improving diagnostic performance of trainees for detecting prostate cancer.

A multicenter study of artificial intelligence-aided software for detecting visible clinically significant prostate cancer on mpMRI

BackgroundAI-based software may improve the performance of radiologists when detecting clinically significant prostate cancer (csPCa). This study aims to compare the performance of radiologists in detecting MRI-visible csPCa on MRI with and without AI-based software.Materials and methodsIn total, 480 multiparametric MRI (mpMRI) images were retrospectively collected from eleven different MR devices, with 349 csPCa lesions in 180 (37.5%) cases. The csPCa areas were annotated based on pathology. Sixteen radiologists from four hospitals participated in reading. Each radiologist was randomly assigned to 30 cases and diagnosed twice. Half cases were interpreted without AI, and the other half were interpreted with AI. After four weeks, the cases were read again in switched mode. The mean diagnostic performance was compared using sensitivity and specificity on lesion level and patient level. The median reading time and diagnostic confidence were assessed.ResultsOn lesion level, AI-aided improved the sensitivity from 40.1% to 59.0% (18.9% increased; 95% confidence interval (CI) [11.5, 26.1]; p < .001). On patient level, AI-aided improved the specificity from 57.7 to 71.7% (14.0% increase, 95% CI [6.4, 21.4]; p < .001) while preserving the sensitivity (88.3% vs. 93.9%, p = 0.06). AI-aided reduced the median reading time of one case by 56.3% from 423 to 185 s (238-s decrease, 95% CI [219, 260]; p < .001), and the median diagnostic confidence score was increased by 10.3% from 3.9 to 4.3 (0.4-score increase, 95% CI [0.3, 0.5]; p < .001).ConclusionsAI software improves the performance of radiologists by reducing false positive detection of prostate cancer patients and also improving reading times and diagnostic confidence.Clinical relevance statementThis study involves the process of data collection, randomization and crossover reading procedure.Graphical

Segmentation of rectal tumor from multi-parametric MRI images using an attention-based fusion network.

Accurate segmentation of rectal tumors is the most crucial task in determining the stage of rectal cancer and developing suitable therapies. However, complex image backgrounds, irregular edge, and poor contrast hinder the related research. This study presents an attention-based multi-modal fusion module to effectively integrate complementary information from different MRI images and suppress redundancy. In addition, a deep learning-based segmentation model (AF-UNet) is designed to achieve accurate segmentation of rectal tumors. This model takes multi-parametric MRI images as input and effectively integrates the features from different multi-parametric MRI images by embedding the attention fusion module. Finally, three types of MRI images (T2, ADC, DWI) of 250 patients with rectal cancer were collected, with the tumor regions delineated by two oncologists. The experimental results show that the proposed method is superior to the most advanced image segmentation method with a Dice coefficient of [Formula: see text], which is also better than other multi-modal fusion methods. Framework of the AF-UNet. This model takes multi-modal MRI images as input, and integrates complementary information using attention mechanism and suppresses redundancy.

Cross-Modality Image Registration Using a Training-Time Privileged Third Modality.

In this work, we consider the task of pairwise cross-modality image registration, which may benefit from exploiting additional images available only at training time from an additional modality that is different to those being registered. As an example, we focus on aligning intra-subject multiparametric Magnetic Resonance (mpMR) images, between T2-weighted (T2w) scans and diffusion-weighted scans with high b-value (DWI [Formula: see text]). For the application of localising tumours in mpMR images, diffusion scans with zero b-value (DWI [Formula: see text]) are considered easier to register to T2w due to the availability of corresponding features. We propose a learning from privileged modality algorithm, using a training-only imaging modality DWI [Formula: see text], to support the challenging multi-modality registration problems. We present experimental results based on 369 sets of 3D multiparametric MRI images from 356 prostate cancer patients and report, with statistical significance, a lowered median target registration error of 4.34 mm, when registering the holdout DWI [Formula: see text] and T2w image pairs, compared with that of 7.96 mm before registration. Results also show that the proposed learning-based registration networks enabled efficient registration with comparable or better accuracy, compared with a classical iterative algorithm and other tested learning-based methods with/without the additional modality. These compared algorithms also failed to produce any significantly improved alignment between DWI [Formula: see text] and T2w in this challenging application.

Performance of Multiparametric MRI and PET Imaging in Identifying and Localizing Locally Radiorecurrent Prostate Cancer in a Phase I/II Trial of Focal Salvage HDR

<h3>Purpose/Objective(s)</h3> Salvage local therapy (SLT) with brachytherapy or SBRT for local radiorecurrence (LRR) of prostate cancer after prior radiotherapy (RT) is increasingly being used due to growing data demonstrating the safety and efficacy of SLT, as well as the availability of advanced imaging to identify LRR, such as multiparametric MRI (mpMRI) and molecular PET imaging (18F-fluciclovine or 18F-DCFPyL PSMA). Further, advanced imaging has led to investigation into focal SLT to improve the therapeutic ratio of reirradiation. Yet, data are limited on the ability of mpMRI and PET to detect LRR and how well they localize the area(s) of recurrence in the prostate to guide focal SLT. Our objective was to determine the ability of mpMRI and PET imaging to detect the presence and full extent of the areas of involvement in men with LRR. <h3>Materials/Methods</h3> We conducted a cohort study of men with LRR of prostate cancer enrolled in the FSHARP phase I/II trial of salvage focal HDR brachytherapy at our institution. Workup included PET/CT, mpMRI of the prostate, and a biopsy documenting the LRR prior to enrollment. Descriptive statistics were used to compare mpMRI and PET detection and localization with biopsy findings to determine the sensitivity/false negative rate (FNR) of each imaging modality, and the frequency with which biopsy reveals disease that would have been missed if the imaging abnormality alone was used to define the focal salvage HDR target. <h3>Results</h3> A total of 40 men were included in this analysis. Median age was 70.5 yrs (Range: 58-83). Median time from initial RT was 9.5 yrs (2-28). Initial NCCN risk groups were as follow: 13 had low risk, 21 had intermediate risk and 2 had high risk. 26 were treated with conventionally fractionated photon RT, 7 with LDR brachytherapy, 4 with protons, 2 with SBRT and 1 with neutron therapy. Regarding recurrence biopsy, 24 men had MRI/TRUS fusion biopsy and 16 had TRUS biopsy alone. A median of 14 cores were obtained and a median of 6 cores were positive. Gleason grade grouping (GG) was as follows: 1 had GG 1, 11 had GG 2, 12 had GG 3, 6 had GG 4, and 4 had GG 5. The median number of MRI target lesions was 1 (0-4), and the median number of lesions on PET was 1 (0-2). 38 men had fluciclovine and 2 had PSMA PET. The table depicts the sensitivity and FNR of each imaging modality, and frequency of pathologically identified disease outside of the imaging target. <h3>Conclusion</h3> Our study emphasizes that though mpMRI and PET are valuable tools for restaging and target volume delineation in LRR of prostate cancer, there are limitations to their use. These imaging modalities cannot be relied upon alone if pursuing focal salvage due to a high FNR, and because the full extent of disease is rarely identified, leading to the risk of marginal misses.

Predicting IDH subtype of grade 4 astrocytoma and glioblastoma from tumor radiomic patterns extracted from multiparametric magnetic resonance images using a machine learning approach.

Background and purposeSemantic imaging features have been used for molecular subclassification of high-grade gliomas. Radiomics-based prediction of molecular subgroups has the potential to strategize and individualize therapy. Using MRI texture features, we propose to distinguish between IDH wild type and IDH mutant type high grade gliomas.MethodsBetween 2013 and 2020, 100 patients were retrospectively analyzed for the radiomics study. Immunohistochemistry of the pathological specimen was used to initially identify patients for the IDH mutant/wild phenotype and was then confirmed by Sanger’s sequencing. Image texture analysis was performed on contrast-enhanced T1 (T1C) and T2 weighted (T2W) MR images. Manual segmentation was performed on MR image slices followed by single-slice multiple sampling image augmentation. Both whole tumor multislice segmentation and single-slice multiple sampling approaches were used to arrive at the best model. Radiomic features were extracted, which included first-order features, second-order (GLCM—Grey level co-occurrence matrix), and shape features. Feature enrichment was done using LASSO (Least Absolute Shrinkage and Selection Operator) regression, followed by radiomic classification using Support Vector Machine (SVM) and a 10-fold cross-validation strategy for model development. The area under the Receiver Operator Characteristic (ROC) curve and predictive accuracy were used as diagnostic metrics to evaluate the model to classify IDH mutant and wild-type subgroups.ResultsMultislice analysis resulted in a better model compared to the single-slice multiple-sampling approach. A total of 164 MR-based texture features were extracted, out of which LASSO regression identified 14 distinctive GLCM features for the endpoint, which were used for further model development. The best model was achieved by using combined T1C and T2W MR images using a Quadratic Support Vector Machine Classifier and a 10-fold internal cross-validation approach, which demonstrated a predictive accuracy of 89% with an AUC of 0.89 for each IDH mutant and IDH wild subgroup.ConclusionA machine learning classifier of radiomic features extracted from multiparametric MRI images (T1C and T2w) provides important diagnostic information for the non-invasive prediction of the IDH mutant or wild-type phenotype of high-grade gliomas and may have potential use in either escalating or de-escalating adjuvant therapy for gliomas or for using targeted agents in the future.

Combination of PI-RADS score and mRNA urine test-A novel scoring system for improved detection of prostate cancer.

Available tests to detect clinically significant prostate cancer frequently lead to overdiagnosis and overtreatment. Our study assessed the feasibility of combining a urinary biomarker-based risk score (SelectMDx®) and multiparametric MRI outcomes in order to identify patients with prostate cancer on prostate biopsy with increased accuracy and reliability. Samples of 74 men with suspicion of prostate cancer and available multiparametric MRI were analysed in a prospective cross-sectional study design. First-voided urine for determination of HOXC6 and DLX1 mRNA levels was collected after digital rectal examination and prior to MRI/ultrasound fusion-guided prostate biopsy. All multiparametric MRI images were centrally reviewed by two experienced radiologists blinded for urine test results and biopsy outcome. The PI-RADS v2 was used. SelectMDx® score, PI-RADS and Gleason Sore were obtained. Associations between Gleason Score, PI-RADS scores and SelectMDx® were assessed using ANOVA and t-test. Sensitivity and specificity were assessed and evaluated as area-under-the-curve of the receiver operating characteristic. Upon biopsy, 59.5% of patients were diagnosed with prostate cancer, whereby 40.6% had high-grade prostate cancer (GS ≥ 7a). SelectMDx® scores were significantly higher for patients with positive biopsy findings (49.07 ± 25.99% vs. 22.00 ± 26.43%; p < 0.001). SelectMDx® scores increased with higher PI-RADS scores. Combining SelectMDx®, history of prior biopsy with benign histology and PI-RADS scores into a novel scoring system led to significant prostate cancer detection rates with tiered detection rate of 39%, 58%, 81% and 100% for Gleason grade group II, III, IV, and V, respectively. The area-under-the-curve for our novel sum score in receiver operating characteristic analysis was 0.84. The synergistic combination of two non-invasive tests into a sum score with increased sensitivity may help avoiding unnecessary biopsies for initial prostate cancer diagnosis. For confirmation, further prospective studies with larger sample sizes and univariate and multivariate regression analyses and decision curve analyses are required.

Prediction Model of Hemorrhage Transformation in Patient with Acute Ischemic Stroke Based on Multiparametric MRI Radiomics and Machine Learning.

Intravenous thrombolysis is the most commonly used drug therapy for patients with acute ischemic stroke, which is often accompanied by complications of intracerebral hemorrhage transformation (HT). This study proposed to build a reliable model for pretreatment prediction of HT. Specifically, 5400 radiomics features were extracted from 20 regions of interest (ROIs) of multiparametric MRI images of 71 patients. Furthermore, a minimal set of all-relevant features were selected by LASSO from all ROIs and used to build a radiomics model through the random forest (RF). To explore the significance of normal ROIs, we built a model only based on abnormal ROIs. In addition, a model combining clinical factors and radiomics features was further built. Finally, the models were tested on an independent validation cohort. The radiomics model with 14 All-ROIs features achieved pretreatment prediction of HT (AUC = 0.871, accuracy = 0.848), which significantly outperformed the model with only 14 Abnormal-ROIs features (AUC = 0.831, accuracy = 0.818). Besides, combining clinical factors with radiomics features further benefited the prediction performance (AUC = 0.911, accuracy = 0.894). So, we think that the combined model can greatly assist doctors in diagnosis. Furthermore, we find that even if there were no lesions in the normal ROIs, they also provide characteristic information for the prediction of HT.