Deep learning CT reconstruction improves liver metastases detection.

OBJECTIVES: Detection of liver metastases is crucial for guiding oncological management. Computed tomography through iterative reconstructions is widely used in this indication but has certain limitations. Deep learning image reconstructions (DLIR) use deep neural networks to achieve a significant noise reduction compared to iterative reconstructions. While reports have demonstrated improvements in image quality, their impact on liver metastases detection remains unclear. Our main objective was to determine whether DLIR affects the number of detected liver metastasis. Our secondary objective was to compare metastases conspicuity between the two reconstruction methods. METHODS: CT images of 121 patients with liver metastases were reconstructed using a 50% adaptive statistical iterative reconstruction (50%-ASiR-V), and three levels of DLIR (DLIR-low, DLIR-medium, and DLIR-high). For each reconstruction, two double-blinded radiologists counted up to a maximum of ten metastases. Visibility and contour definitions were also assessed. Comparisons between methods for continuous parameters were performed using mixed models. RESULTS: A higher number of metastases was detected by one reader with DLIR-high: 7 (2-10) (median (Q1-Q3); total 733) versus 5 (2-10), respectively for DLIR-medium, DLIR-low, and ASiR-V (p < 0.001). Ten patents were detected with more metastases with DLIR-high simultaneously by both readers and a third reader for confirmation. Metastases visibility and contour definition were better with DLIR than ASiR-V. CONCLUSION: DLIR-high enhanced the detection and visibility of liver metastases compared to ASiR-V, and also increased the number of liver metastases detected. CRITICAL RELEVANCE STATEMENT: Deep learning-based reconstruction at high strength allowed an increase in liver metastases detection compared to hybrid iterative reconstruction and can be used in clinical oncology imaging to help overcome the limitations of CT. KEY POINTS: Detection of liver metastases is crucial but limited with standard CT reconstructions. More liver metastases were detected with deep-learning CT reconstruction compared to iterative reconstruction. Deep learning reconstructions are suitable for hepatic metastases staging and follow-up.

A deep learning-based interactive medical image segmentation framework with sequential memory.

BACKGROUND AND OBJECTIVE: Image segmentation is an essential component in medical image analysis. The case of 3D images such as MRI is particularly challenging and time consuming. Interactive or semi-automatic methods are thus highly desirable. However, existing methods do not exploit the typical sequentiality of real user interactions. This is due to the interaction memory used in these systems, which discards ordering. In contrast, we argue that the order of the user corrections should be used for training and lead to performance improvements. METHODS: We contribute to solving this problem by proposing a general multi-class deep learning-based interactive framework for image segmentation, which embeds a base network in a user interaction loop with a user feedback memory. We propose to model the memory explicitly as a sequence of consecutive system states, from which the features can be learned, generally learning from the segmentation refinement process. Training is a major difficulty owing to the network's input being dependent on the previous output. We adapt the network to this loop by introducing a virtual user in the training process, modelled by dynamically simulating the iterative user feedback. RESULTS: We evaluated our framework against existing methods on the complex task of multi-class semantic instance female pelvis MRI segmentation with 5 classes, including up to 27 tumour instances, using a segmentation dataset collected in our hospital, and on liver and pancreas CT segmentation, using public datasets. We conducted a user evaluation, involving both senior and junior medical personnel in matching and adjacent areas of expertise. We observed an annotation time reduction with 5'56" for our framework against 25' on average for classical tools. We systematically evaluated the influence of the number of clicks on the segmentation accuracy. A single interaction round our framework outperforms existing automatic systems with a comparable setup. We provide an ablation study and show that our framework outperforms existing interactive systems. CONCLUSIONS: Our framework largely outperforms existing systems in accuracy, with the largest impact on the smallest, most difficult classes, and drastically reduces the average user segmentation time with fast inference at 47.2±6.2 ms per image.

Multicenter External Validation of the Deep Pelvic Endometriosis Index Magnetic Resonance Imaging Score.

Importance: Preoperative mapping of deep pelvic endometriosis (DPE) is crucial as surgery can be complex and the quality of preoperative information is key. Objective: To evaluate the Deep Pelvic Endometriosis Index (dPEI) magnetic resonance imaging (MRI) score in a multicenter cohort. Design, Setting, and Participants: In this cohort study, the surgical databases of 7 French referral centers were retrospectively queried for women who underwent surgery and preoperative MRI for DPE between January 1, 2019, and December 31, 2020. Data were analyzed in October 2022. Intervention: Magnetic resonance imaging scans were reviewed using a dedicated lexicon and classified according to the dPEI score. Main outcomes and measures: Operating time, hospital stay, Clavien-Dindo-graded postoperative complications, and presence of de novo voiding dysfunction. Results: The final cohort consisted of 605 women (mean age, 33.3; 95% CI, 32.7-33.8 years). A mild dPEI score was reported in 61.2% (370) of the women, moderate in 25.8% (156), and severe in 13.1% (79). Central endometriosis was described in 93.2% (564) of the women and lateral endometriosis in 31.2% (189). Lateral endometriosis was more frequent in severe (98.7%) vs moderate (48.7%) disease and in moderate vs mild (6.7%) disease according to the dPEI (P < .001). Median operating time (211 minutes) and hospital stay (6 days) were longer in severe DPE than in moderate DPE (operating time, 150 minutes; hospital stay 4 days; P < .001), and in moderate than in mild DPE (operating time; 110 minutes; hospital stay, 3 days; P < .001). Patients with severe disease were 3.6 times more likely to experience severe complications than patients with mild or moderate disease (odds ratio [OR], 3.6; 95% CI, 1.4-8.9; P = .004). They were also more likely to experience postoperative voiding dysfunction (OR, 3.5; 95% CI, 1.6-7.6; P = .001). Interobserver agreement between senior and junior readers was good (κ = 0.76; 95% CI, 0.65-0.86). Conclusions and Relevance: The findings of this study suggest the ability of the dPEI to predict operating time, hospital stay, postoperative complications, and de novo postoperative voiding dysfunction in a multicenter cohort. The dPEI may help clinicians to better anticipate the extent of DPE and improve clinical management and patient counseling.

Metastatic melanoma treated by immunotherapy: discovering prognostic markers from radiomics analysis of pretreatment CT with feature selection and classification.

PURPOSE: Immunotherapy has dramatically improved the prognosis of patients with metastatic melanoma (MM). Yet, there is a lack of biomarkers to predict whether a patient will benefit from immunotherapy. Our aim was to create radiomics models on pretreatment computed tomography (CT) to predict overall survival (OS) and treatment response in patients with MM treated with anti-PD-1 immunotherapy. METHODS: We performed a monocentric retrospective analysis of 503 metastatic lesions in 71 patients with 46 radiomics features extracted following lesion segmentation. Predictive accuracies for OS < 1 year versus > 1 year and treatment response versus no response was compared for five feature selection methods (sequential forward selection, recursive, Boruta, relief, random forest) and four classifiers (support vector machine (SVM), random forest, K-nearest neighbor, logistic regression (LR)) used with or without SMOTE data augmentation. A fivefold cross-validation was performed at the patient level, with a tumour-based classification. RESULTS: The highest accuracy level for OS predictions was obtained with 3D lesions (0.91) without clinical data integration when combining Boruta feature selection and the LR classifier, The highest accuracy for treatment response prediction was obtained with 3D lesions (0.88) without clinical data integration when combining Boruta feature selection, the LR classifier and SMOTE data augmentation. The accuracy was significantly higher concerning OS prediction with 3D segmentation (0.91 vs 0.86) while clinical data integration led to improved accuracy notably in 2D lesions (0.76 vs 0.87) regarding treatment response prediction. Skewness was the only feature found to be an independent predictor of OS (HR (CI 95%) 1.34, p-value 0.001). CONCLUSION: This is the first study to investigate CT texture parameter selection and classification methods for predicting MM prognosis with treatment by immunotherapy. Combining pretreatment CT radiomics features from a single tumor with data selection and classifiers may accurately predict OS and treatment response in MM treated with anti-PD-1.

Protocol for the e-POWUS Project: multicentre blinded-randomised controlled trial of ultrasound speed choice to improve sonography quality in pregnant women with obesity.

INTRODUCTION: During pregnancy, maternal obesity increases the risk of fetal abnormalities. Despite advances in ultrasound imaging, the assessment of fetal anatomy is less thorough among these women. Currently, the construction of ultrasound images uses a conventional ultrasound propagation velocity (1540 m/s), which does not correspond to the slower speed of propagation in fat tissue.The main objective of this randomised study is to compare the completeness of fetal ultrasonography according to whether the operator could choose the ultrasound velocity (1420, 1480 or 1540 m/s) or was required to apply the 1540 m/s velocity. METHODS AND ANALYSIS: This randomised trial is an impact study to compare a diagnostic innovation with the reference technique. The trial inclusion criteria require that a pregnant woman with obesity be undergoing a fetal morphology examination by ultrasound from 20+0 to 25+0 gestational weeks.Randomisation will allocate women into two groups. The first will be the 'modulable speed' group, in which operators can choose the speed of ultrasound propagation to be considered for the morphological analysis: 1420, 1480 or 1540 m/s. In the second 'conventional speed' group, operators will perform the morphological examination with the ultrasound speed fixed at 1540 m/s. The adjudication committee, two independent experts, will validate the completeness of each examination and the quality of the images. ETHICS AND DISSEMINATION: This research protocol does not change the standard management. The only possible impact is an improvement of the ultrasound examination by improving the quality of the image and the completeness of morphological examination. The Agence du Médicament et produits de santé approved this study (2018-A03478-47). The anonymised data will be available on request from the principal investigator. Results will be reported in peer-reviewed journals and at scientific meetings. TRIAL REGISTRATION NUMBER: ClinicalTrials.gov (http://www.clinicaltrials.gov) Registry (NCT04212234).

Quantification of steatosis in alcoholic and nonalcoholic fatty liver disease: Evaluation of four MR techniques versus biopsy.

PURPOSE: Given the growing prevalence of obesity and metabolic syndrome, the management of hepatic steatosis, especially its quantification, is a major issue. We assessed the quantification of liver steatosis using four different MR methods, in order to determine the one that is best correlated with the reference method which consists of histological measurement by liver biopsy. METHOD: Seventy-one successive patients requiring liver biopsy for acute or chronic liver disease were enrolled prospectively between March 2017 and March 2018, 11 were excluded and 60 were reported. Liver MR (1.5 T) was organised in order to be performed the same day, using four different steatosis quantification techniques (3-echo MRI, 6-echo MRI, 11-echo MRI and MR Spectroscopy). Quantitative histological and imaging data were compared. In a secondary analysis, we studied the possible influence of alcohol drinking, hepatic iron overload, and the presence of liver fibrosis. RESULTS: All four MR techniques were found to have excellent correlations with the histological measurements: 3-echo MRI (r = 0.852, p < 0.001), 6-echo MRI (r = 0.819, p < 0.001), 11-echo MRI (r = 0.818, p < 0.001) and MR Spectroscopy (r = 0,812, p < 0,001). Interestingly, we also found that the presence of alcohol consumption, iron overload and fibrosis did not interfere with measurements, whichever technique was used. CONCLUSION: In the evaluation of hepatic steatosis, our study showed very good correlations of all four MR techniques with the histological standard. There was no confounding factor in a representative group of patients with associated liver conditions such as alcohol consumption, fibrosis and iron overload, for each technique. All four MR techniques may be used in daily practice.