Prostate cancer is the second leading cause of death among men in United States. The diagnosis prostate MRI often relies on accurate zonal segmentation. However, state-of-the-art automatic segmentation methods fail to produce well-contained volumetric zones since certain slices MRI, such as base and apex slices, are harder segment than other slices. This difficulty can be overcome by leveraging important multi-scale image-based information from adjacent but current do not fully learn exploit...

Conditional image generation plays a vital role in medical analysis as it is effective tasks such super-resolution, denoising, and inpainting, among others. Diffusion models have been shown to perform at state-of-the-art level natural generation, but they not thoroughly studied with specific conditions. Moreover, current their own problems, limiting usage various tasks. In this paper, we introduce the use of conditional Denoising Probabilistic Models (cDDPMs) for which achieve performance on several

The current standardized scheme for interpreting MRI requires a high level of expertise and exhibits significant degree inter-reader intra-reader variability. An automated prostate cancer (PCa) classification can improve the ability to assess spectrum PCa. purpose study was evaluate performance texture-based deep learning model (Textured-DL) differentiating between clinically PCa (csPCa) non-csPCa compare Textured-DL with Prostate Imaging Reporting Data System (PI-RADS)-based (PI-RADS-CLA),...

Diffusion models have achieved impressive performance on various image generation tasks, including super-resolution. Despite their performance, diffusion suffer from high computational costs due to the large number of denoising steps. In this paper, we proposed a novel accelerated model, termed Partial Models (PDMs), for magnetic resonance imaging (MRI) We observed that latents diffusing pair low- and high-resolution images gradually converge become indistinguishable after certain noise...

Abstract Multi-parametric MRI (mpMRI) is widely used for prostate cancer (PCa) diagnosis. Deep learning models show good performance in detecting PCa on mpMRI, but domain-specific PCa-related anatomical information sometimes overlooked and not fully explored even by state-of-the-art deep models, causing potential suboptimal performances detection. Symmetric-related commonly when distinguishing lesions from other visually similar benign tissue. In addition, different combinations of mpMRI...

A large portion of volumetric medical data, especially magnetic resonance imaging (MRI) is anisotropic, as the through-plane resolution typically much lower than in-plane resolution. Both 3D and purely 2D deep learning-based segmentation methods are deficient in dealing with such data since performance suffers when confronting anisotropic disregard crucial information. Insufficient work has been done on 2.5D methods, which convolution mainly used concert These models focus learning...

Multiparametric MRI (mpMRI) is commonly recommended as a triage test prior to any prostate biopsy. However, there exists limited consensus on which patients with negative mpMRI could avoid To identify patient safely biopsy when the negative, via radiomics-based machine learning approach. Retrospective. Three hundred thirty 3T between January 2016 and December 2018 were included. A 3.0 T/T2-weighted turbo spin echo (TSE) imaging (T2 WI) diffusion-weighted (DWI). The integrative (iML) model...

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) measures microvascular perfusion by capturing the temporal changes of an MRI contrast agent in a target tissue, and it provides valuable information for diagnosis prognosis wide range tumors. Quantitative DCE-MRI analysis commonly relies on nonlinear least square (NLLS) fitting pharmacokinetic (PK) model to concentration curves. However, voxel-wise application such curve is highly time-consuming. The arterial input function (AIF)...

Whole-prostate gland (WPG) segmentation plays a significant role in prostate volume measurement, treatment, and biopsy planning. This study evaluated previously developed automatic WPG segmentation, deep attentive neural network (DANN), on large, continuous patient cohort to test its feasibility clinical setting. With IRB approval HIPAA compliance, the included 3,698 3T MRI scans acquired between 2016 2020. In total, 335 were used train model, 3,210 100 conduct qualitative quantitative...

Motivation: Fully automatic segmentation of fibroglandular tissue (FGT) and background parenchymal enhancement (BPE) quantification methods with high generalizability for different FGT levels are still lacking. Goal(s): We aimed to improve the accuracy across various that accurately quantify density BPE. Approach: A novel anatomy-aware loss function based on variations in level was applied a fully model training breast MRIs. Results: The segmentation, estimation, BPE were improved at levels....

Denoising diffusion probabilistic models (DDPMs) have achieved impressive performance on various image generation tasks, including super-resolution. By learning to reverse the process of gradually diffusing data distribution into Gaussian noise, DDPMs generate new by iteratively denoising from random noise. Despite their performance, diffusion-based generative suffer high computational costs due large number steps.In this paper, we first observed that intermediate latent states converge and...

Current deep learning-based models typically analyze medical images in either 2D or 3D albeit disregarding volumetric information suffering sub-optimal performance due to the anisotropic resolution of MR data. Furthermore, providing an accurate uncertainty estimation is beneficial clinicians, as it indicates how confident a model about its prediction. We propose novel 2.5D cross-slice attention that utilizes both global and local information, along with evidential critical loss, perform...

A deep learning based DCE-MRI analysis method was proposed with a dedicated neural network architecture and data generation framework. The does not need acquisition or annotation for training. Compared to conventional non-linear least square (NLLS) fitting methods, the significantly reduced average processing time from hours few minutes while preserved estimation quality.

The study aimed to build a deep-learning-based prostate cancer (PCa) detection model integrating the anatomical priors related PCa’s zonal appearance difference and asymmetric patterns of PCa. A total 220 patients with 246 whole-mount histopathology (WMHP) confirmed clinically significant (csPCa), 432 no indication lesions on multi-parametric MRI (mpMRI) were included in study. proposed 3D Siamese nnUNet self-designed Zonal Loss was implemented, results evaluated using 5-fold...

Motivation: Despite the growing use of multiparametric MRI (mpMRI), there remains an unmet need for additional quantitative methods to improve prostate cancer (PCa) localization by anatomic zones. Goal(s): To extract radiomics features that determine differences in detection rates (DRs) and positive predictive values (PPV) clinically significant PCa (csPCa). Approach: We extracted shape- first-order based from 543 csPCa lesions across 468 male subjects used Mann-Whitney U test assess key...

Multi-parametric MRI (mpMRI) is a powerful non-invasive tool for diagnosing prostate cancer (PCa) and widely recommended to be performed before biopsies. Prostate Imaging Reporting Data System version (PI-RADS) used interpret mpMRI. However, when the pre-biopsy mpMRI negative, PI-RADS 1 or 2, there exists no consensus on which patients should undergo Recently, radiomics has shown great abilities in quantitative imaging analysis with outstanding performance computer-aid diagnosis tasks. We...

Whole-prostate gland (WPG) segmentation plays a significant role in prostate volume measurement, treatment, and biopsy planning. This study evaluated previously developed automatic WPG segmentation, deep attentive neural network (DANN), on large, continuous patient cohort to test its feasibility clinical setting.

A large portion of volumetric medical data, especially magnetic resonance imaging (MRI) is anisotropic, as the through-plane resolution typically much lower than in-plane resolution. Both 3D and purely 2D deep learning-based segmentation methods are deficient in dealing with such data since performance suffers when confronting anisotropic disregard crucial information. Insufficient work has been done on 2.5D methods, which convolution mainly used concert These models focus learning...