Semantic Instance Segmentation of Kidney Cysts in MR Images: A Fully Automated 3D Approach Developed Through Active Learning.

Total kidney volume (TKV) is the main imaging biomarker used to monitor disease progression and to classify patients affected by autosomal dominant polycystic kidney disease (ADPKD) for clinical trials. However, patients with similar TKVs may have drastically different cystic presentations and phenotypes. In an effort to quantify these cystic differences, we developed the first 3D semantic instance cyst segmentation algorithm for kidneys in MR images. We have reformulated both the object detection/localization task and the instance-based segmentation task into a semantic segmentation task. This allowed us to solve this unique imaging problem efficiently, even for patients with thousands of cysts. To do this, a convolutional neural network (CNN) was trained to learn cyst edges and cyst cores. Images were converted from instance cyst segmentations to semantic edge-core segmentations by applying a 3D erosion morphology operator to up-sampled versions of the images. The reduced cysts were labeled as core; the eroded areas were dilated in 2D and labeled as edge. The network was trained on 30 MR images and validated on 10 MR images using a fourfold cross-validation procedure. The final ensemble model was tested on 20 MR images not seen during the initial training/validation. The results from the test set were compared to segmentations from two readers. The presented model achieved an averaged R2 value of 0.94 for cyst count, 1.00 for total cyst volume, 0.94 for cystic index, and an averaged Dice coefficient of 0.85. These results demonstrate the feasibility of performing cyst segmentations automatically in ADPKD patients.

Automatic semantic segmentation of kidney cysts in MR images of patients affected by autosomal-dominant polycystic kidney disease.

PURPOSE: For patients affected by autosomal-dominant polycystic kidney disease (ADPKD), successful differentiation of cysts is useful for automatic classification of patient phenotypes, clinical decision-making, and disease progression. The objective was to develop and evaluate a fully automated semantic segmentation method to differentiate and analyze renal cysts in patients with ADPKD. METHODS: An automated deep learning approach using a convolutional neural network was trained, validated, and tested on a set of 60 MR T2-weighted images. A three-fold cross-validation approach was used to train three models on distinct training and validation sets (n = 40). An ensemble model was then built and tested on the hold out cases (n = 20), with each of the cases compared to manual segmentations performed by two readers. Segmentation agreement between readers and the automated method was assessed. RESULTS: The automated approach was found to perform at the level of interobserver variability. The automated approach had a Dice coefficient (mean ± standard deviation) of 0.86 ± 0.10 vs Reader-1 and 0.84 ± 0.11 vs. Reader-2. Interobserver Dice was 0.86 ± 0.08. In terms of total cyst volume (TCV), the automated approach had a percent difference of 3.9 ± 19.1% vs Reader-1 and 8.0 ± 24.1% vs Reader-2, whereas interobserver variability was - 2.0 ± 16.4%. CONCLUSION: This study developed and validated a fully automated approach for performing semantic segmentation of kidney cysts in MR images of patients affected by ADPKD. This approach will be useful for exploring additional imaging biomarkers of ADPKD and automatically classifying phenotypes.

Automated total kidney volume measurements in pre-clinical magnetic resonance imaging for resourcing imaging data, annotations, and source code.

The objective of this study was to validate a fully automated total kidney volume measurement method for pre-clinical rodent trials that is fast, accurate, reproducible, and to provide these resources to the research community. Rodent studies that involve imaging are crucial for monitoring treatment efficacy in diseases such as polycystic kidney disease. Previous studies utilize manual or semi-automated segmentations, which are time consuming and potentially biased. To develop our automated system, a total of 150 axial magnetic resonance images (MRI) from a variety of mouse models were manually segmented and used to train/validate an automated algorithm. To test the longitudinal application of the model, four mutant and four wild-type mice were imaged sequentially over three to twelve weeks via MRI. Segmentations of the kidneys (excluding the renal pelvis) were generated by the automated method and two different readers, with one reader repeating the measurements. Similarity metrics and longitudinal analysis were calculated to assess the performance of the automated compared to the manual methods. The automated approach required no user input, besides a final visual quality control step. Similarity metrics of the automated method versus the manual segmentations were on par with inter- and intra-reader comparisons. Thus, our fully automated approach described here can be safely used in longitudinal, pre-clinical trials that involve the segmentation of rodent kidneys in T2-weighted MRIs.

Borderline Resectable and Locally Advanced Pancreatic Cancer: FDG PET/MRI and CT Tumor Metrics for Assessment of Pathologic Response to Neoadjuvant Therapy and Prediction of Survival.

BACKGROUND. Imaging biomarkers of response to neoadjuvant therapy (NAT) for pancreatic ductal adenocarcinoma (PDA) are needed to optimize treatment decisions and long-term outcomes. OBJECTIVE. The purpose of this study was to investigate metrics from PET/MRI and CT to assess pathologic response of PDA to NAT and to predict overall survival (OS). METHODS. This retrospective study included 44 patients with 18F-FDG-avid borderline resectable or locally advanced PDA on pretreatment PET/MRI who also underwent post-NAT PET/MRI before surgery between August 2016 and February 2019. Carbohydrate antigen 19-9 (CA 19-9) level, metabolic metrics from PET/MRI, and morphologic metrics from CT (n = 34) were compared between pathologic responders (College of American Pathologists scores 0 and 1) and nonresponders (scores 2 and 3). AUCs were measured for metrics significantly associated with pathologic response. Relation to OS was evaluated with Cox proportional hazards models. RESULTS. Among 44 patients (22 men, 22 women; mean age, 62 ± 11.6 years), 19 (43%) were responders, and 25 (57%) were nonresponders. Median OS was 24 months (range, 6-42 months). Before treatment, responders and nonresponders did not differ in CA 19-9 level, metabolic metrics, or CT metrics (p > .05). After treatment, responders and nonresponders differed in complete metabolic response (CMR) (responders, 89% [17/19]; nonresponders, 40% [10/25]; p = .04], mean change in SUVmax (ΔSUVmax; responders, -70% ± 13%; nonresponders, -37% ± 42%; p < .001), mean change in SUVmax corrected to serum glucose level (ΔSUVgluc) (responders, -74% ± 12%; nonresponders, -30% ± 58%; p < .001), RECIST response on CT (responders, 93% [13/14]; nonresponders, 50% [10/20]; p = .02)], and mean change in tumor volume on CT (ΔTvol) (responders, -85% ± 21%; nonresponders, 57% ± 400%; p < .001). The AUC of CMR for pathologic response was 0.75; ΔSUVmax, 0.83; ΔSUVgluc, 0.87; RECIST, 0.71; and ΔTvol 0.86. The AUCs of bivariable PET/MRI and CT models were 0.83 (CMR and ΔSUVmax), 0.87 (CMR and ΔSUVgluc), and 0.87 (RECIST and ΔTvol). OS was associated with CMR (p = .03), ΔSUVmax (p = .003), ΔSUVgluc (p = .003), and RECIST (p = .046). CONCLUSION. Unlike CA 19-9 level, changes in metabolic metrics from PET/MRI and morphologic metrics from CT after NAT were associated with pathologic response and OS in patients with PDA, warranting prospective validation. CLINICAL IMPACT. Imaging metrics associated with pathologic response and OS in PDA could help guide clinical management and outcomes for patients with PDA who undergo emergency therapeutic interventions.