MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541122835 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Dec. 5, 2025, for 'hybrid transformer-cnn architecture for pneumothorax segmentation in chest radiographs.'
Inventor(s) include Dr. Umamaheswa RI M; Kousik C; and Neeshanth S. M.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure provides a system for automated pneumothorax segmentation in chest radiographs. The system includes a convolutional neural network backbone (ResNet50) configured to extract hierarchical feature representations from input chest radiograph images, a transformer encoder-decoder architecture configured to process the hierarchical feature representations and model global context through attention mechanisms, a boundary supervision mechanism configured to process ground truth segmentation masks to generate supervisory signals identifying regions of boundary ambiguity, boundary-aware attention mechanisms within the transformer encoder-decoder architecture configured to focus processing on regions identified by the boundary supervision mechanism, a fusion mechanism configured to combine features from the convolutional neural network backbone with transformer-processed features, and a segmentation output mechanism configured to generate predicted segmentation maps for pneumothorax regions. The boundary supervision mechanism applies distance transform and local maxima detection to generate key patch maps guiding the boundary-aware attention gates."
Disclaimer: Curated by HT Syndication.
