Category Ranking

98%

Total Visits

921

Avg Visit Duration

2 minutes

Citations

20

Article Abstract

This work introduces T-Loss, a novel and robust loss function for medical image segmentation. T-Loss is derived from the negative log-likelihood of the Student-t distribution and excels at handling noisy masks by dynamically controlling its sensitivity through a single parameter. This parameter is optimized during the backpropagation process, obviating the need for additional computations or prior knowledge about the extent and distribution of noisy labels. We provide in-depth analysis of this parameter behavior during training and revealing its adaptive nature and its role in preventing noisy memorization. Our extensive experiments demonstrate that T-Loss significantly outperforms traditional loss functions in terms of dice scores on two public medical datasets, specifically for skin lesion and lung segmentation. Moreover, T-Loss exhibits remarkable resilience to various types of simulated label noise, which mimics human annotation errors. Our results provide strong evidence that T-Loss is a promising alternative for medical image segmentation where high levels of noise or outliers in the dataset are a typical phenomenon in practice. The project website, including code and additional resources, can be found at: https://robust-tloss.github.io/.

Download full-text PDF

Source
http://dx.doi.org/10.1016/j.media.2025.103735DOI Listing

Publication Analysis

Top Keywords

medical image
12
image segmentation
12
segmentation t-loss
8
t-loss
5
robust t-loss
4
medical
4
t-loss medical
4
segmentation
4
segmentation work
4
work introduces
4

Similar Publications

Introduction: Medical physicists play a critical role in ensuring image quality and patient safety, but their routine evaluations are limited in scope and frequency compared to the breadth of clinical imaging practices. An electronic radiologist feedback system can augment medical physics oversight for quality improvement. This work presents a novel quality feedback system integrated into the Epic electronic medical record (EMR) at a university hospital system, designed to facilitate feedback from radiologists to medical physicists and technologist leaders.

View Article and Find Full Text PDF

Eating disorders are primarily associated with women and an obsession with thinness. Recent research and social media content show that men are also concerned about their body image, striving for a muscular and athletic physique. To investigate eating disorder tendencies among male content creators with a mesomorphic body type (N = 26), a social media analysis was conducted on Instagram and TikTok over four weeks.

View Article and Find Full Text PDF

Background: This study aimed to investigate the gender-specific associations of skeletal muscle mass and fat mass with non-alcoholic fatty liver disease (NAFLD) and NAFLD-related liver fibrosis in two population-based studies.

Methods: Analyses were based on data from the MEGA (n = 238) and the MEIA study (n = 594) conducted between 2018 and 2023 in Augsburg, Germany. Bioelectrical impedance analysis was used to evaluate relative skeletal muscle mass (rSM) and SM index (SMI) as well as relative fat mass (rFM) and FM index (FMI); furthermore, the fat-to-muscle ratio was built.

View Article and Find Full Text PDF

MRI-negative cerebellar syndrome caused by medication-induced magnesium deficiency: a case report.

BMC Neurol

September 2025

Department of Neurology, University Hospital, RWTH Aachen University, Pauwelsstrasse 30, Aachen, North Rhine-Westphalia, Germany.

Background: Cerebellar pathologies in adults can have a wide range of hereditary, acquired and sporadic-degenerative causes. Due to the frequency in daily hospital, especially intensive care, settings, electrolyte imbalances are an important, yet rare differential diagnosis. The hypomagnesemia-induced cerebellar syndrome (HiCS) constitutes a relevant disease entity with clinical and morphological variability due to a potential progression of symptoms and a promising causal treatment.

View Article and Find Full Text PDF

Background: IgG4-related lung disease (IgG4-RLD) is a rare autoimmune condition. This study aims to systematically analyze the clinical characteristics of IgG4-RLD to enhance clinicians' awareness and improve patient outcomes.

Methods: This retrospective analysis investigates the clinical data of 20 patients diagnosed with IgG4-RLD at the Yichang Central People's Hospital between January 2019 and April 2025.

View Article and Find Full Text PDF