Semi-supervised model based on implicit neural representation and mutual learning (SIMN) for multi-center nasopharyngeal carcinoma segmentation on MRI.

BACKGROUND: The issue of using deep learning to obtain accurate gross tumor volume (GTV) and metastatic lymph nodes (MLN) segmentation for nasopharyngeal carcinoma (NPC) on heterogeneous magnetic resonance imaging (MRI) images with limited labeling remains unsolved. METHOD: We collected 918 patients with MRI images from three hospitals to develop and validate models and proposed a semi-supervised framework for the fine delineation of multi-center NPC boundaries by integrating uncertainty-based implicit neural representations named SIMN. The framework utilizes the deep mutual learning approach with CNN and Transformer, incorporating dynamic thresholds. Additionally, domain adaptive algorithms are employed to enhance the performance. RESULTS: SIMN predictions have a high overlap ratio with the ground truth. Under the 20 % labeled cases, for the internal test cohorts, the average DSC in GTV and MLN are 0.7981 and 0.7804, respectively; for external test cohort Wu Zhou Red Cross Hospital, the average DSC in GTV and MLN are 0.7217 and 0.7581, respectively; for external test cohorts First People Hospital of Foshan, the average DSC in GTV and MLN are 0.7004 and 0.7692, respectively. No significant differences are found in DSC, HD95, ASD, and Recall for patients with different clinical categories. Moreover, SIMN outperformed existing classical semi-supervised methods. CONCLUSIONS: SIMN showed a highly accurate GTV and MLN segmentation for NPC on multi-center MRI images under Semi-Supervised Learning (SSL), which can easily transfer to other centers without fine-tuning. It suggests that it has the potential to act as a generalized delineation solution for heterogeneous MRI images with limited labels in clinical deployment.

Stress-only versus rest-stress SPECT MPI in the detection and diagnosis of myocardial ischemia and infarction by machine learning.

BACKGROUND: Rest-stress SPECT myocardial perfusion imaging (MPI) is widely used to evaluate coronary artery disease (CAD). We aim to evaluate stress-only versus rest-stress MPI in diagnosing CAD by machine learning (ML). METHODS: A total of 276 patients with suspected CAD were randomly divided into training (184 patients) and validation (92 patients) cohorts. Variables extracted from clinical, physiological, and rest-stress SPECT MPI were screened. Stress-only and rest-stress MPI using ML were established and compared using the training cohort. Then the diagnostic performance of two models in diagnosing myocardial ischemia and infarction was evaluated in the validation cohort. RESULTS: Six ML models based on stress-only MPI selected summed stress score, summed wall thickness score of stress%, and end-diastolic volume of stress as key variables and performed equally good as rest-stress MPI in detecting CAD [area under the curve (AUC): 0.863 versus 0.877, P  = 0.519]. Furthermore, stress-only MPI showed a reasonable prediction of reversible deficit, as shown by rest-stress MPI (AUC: 0.861). Subsequently, nomogram models using the above-stated stress-only MPI variables showed a good prediction of CAD and reversible perfusion deficit in training and validation cohorts. CONCLUSION: Stress-only MPI demonstrated similar diagnostic performance compared with rest-stress MPI using 6 ML algorithms. Stress-only MPI with ML models can diagnose CAD and predict ischemia from scar.