Praneeta R. Konduri

All Publications

• Deep generative models for vessel segmentation in CT angiography of the brain. Computers in biology and medicine van Voorst, H., Su, J., Konduri, P. R., Majoie, C. B., Roos, Y. B., Emmer, B. J., Marquering, H. A., de Vos, B. D., Caan, M. W., Isgum, I., MR CLEAN Registry collaborators 2026; 202: 111432

  Abstract

  Automated vessel segmentation in brain CT angiography (CTA) remains challenging despite the potential benefits of its applications. Expert acquisition of reference vessel segmentations is a laborious task. We propose an unsupervised generative deep learning approach that can be trained for vessel segmentation in brain CTA using a large dataset (n=908) of unlabelled brain CTAs and non-contrast enhanced CTs (NCCTs). Our semi-supervised approach uses a conditional generative adversarial network (GAN) for CTA to NCCT translation by generating a contrast map that allows for automatic extraction of vessel segmentations. Furthermore, we propose a 3D Frangi filter-based loss function to enhance tubular structures in the contrast map to improve vessel segmentations. We used a hold-out test set of 9 CTA volumes with manually annotated reference segmentations. We compared our semi-supervised approach with a state-of-the-art supervised nnUnet, trained and evaluated with test set using 9-fold nested cross-validation. Evaluation metrics included voxel-wise Dice similarity coefficient (DSC), true positive rate (TPR), and false positive rate (FPR). The DSC was 4 % lower for the semi-supervised approach (DSC: 0.74) compared to the supervised nnUnet (DSC: 0.78). Both the TPR and FPR were higher for the semi-supervised approach (TPR: 0.75, FPR/1000 voxels:2.05) compared to the supervised nnUnet (TPR:0.71, FPR/1000 voxels:0.87). Hence, the quantitative results showed that our semi-supervised method approaches a supervised state-of-the-art segmentation network. The results demonstrate that a semi-supervised generative deep learning approach for the segmentation of intracranial vessels is feasible without laborious manual segmentations.

  View details for DOI 10.1016/j.compbiomed.2025.111432

  View details for PubMedID 41494368

• Hypodensity Beyond the Ischemic Core: Penumbral Changes Detected With Relative Noncontrast Computed Tomography. Stroke Vandewalle, L., Konduri, P. R., Christensen, S., Seners, P., Wouters, A., Yuen, N., Mlynash, M., Kemp, S., Heit, J. J., Albers, G. W., Demeestere, J., Lansberg, M. G., Lemmens, R. 2025

  Abstract

  BACKGROUND: In acute ischemic stroke, infarcted tissue gradually becomes detectable on noncontrast computed tomography (NCCT) as a hypodensity representing vasogenic edema. We studied whether subtle NCCT density changes are also present in penumbral tissue.METHODS: This observational cohort study included patients with stroke with anterior circulation occlusions from the CRISP2 study (CT Perfusion to Predict Response to Recanalization in Ischemic Stroke Project 2) who were transferred from a primary to a comprehensive stroke center for consideration of endovascular thrombectomy. Patients received baseline NCCT and computed tomography perfusion at the referring hospital and magnetic resonance imaging at the receiving hospital. We created baseline relative NCCT images, which compare voxel density to the corresponding area in the contralateral hemisphere. We analyzed the relative density of rNCCT in the core and penumbra (based on computed tomography perfusion in referring hospitals). We also assessed the correlation between relative density and the degree of hypoperfusion in the penumbra, defined as the time-to-maximum of the tissue residue function. We studied the association between penumbral changes and functional outcomes on the full distribution of the modified Rankin Scale score at 90 days.RESULTS: From the 314 patients, 162 met inclusion criteria with a median (interquartile range) age of 73 (61-83) years, penumbra volume of 78 (52-113) mL, and core volume of 0.6 (0-13.0) mL; 54% were men. The relative density was reduced by a median of 1.8% (P<0.0001) in the penumbra and 3.3% in the core (P<0.0001). Relative hypodensity in the penumbra was more profound with increasing hypoperfusion: 1.5% in regions with time-to-maximum of 6- to 8-s region, 1.8% in time-to-maximum of 8- to 10-s v, and 2.2% in time-to-maximum >10-s region (P<0.0001). We identified a trend toward worse outcomes with more hypodense penumbra (odds ratio, 1.193 [95% CI, 0.996-1.430]).CONCLUSIONS: In patients with anterior circulation acute ischemic stroke, we identified relative hypodensity in penumbral tissue on NCCT with potential clinical relevance on 90-day functional outcomes. The hypodensity was more pronounced with increasing hypoperfusion severity.

  View details for DOI 10.1161/STROKEAHA.124.050317

  View details for PubMedID 40557487