CAHA-Net: A novel MR image classification model based on DenseNet incorporating coordinate attention and hybrid augmentation

Hui Qi, Ruizhe Guo, Jiajun He, Xiyi Wang, Yaling Han, Xuchu Jiang, Chibiao Liu · 2026

Read CAHA-Net: A novel MR image classification model based on DenseNet incorporating coordinate attention and hybrid augmentation by Hui Qi, Ruizhe Guo, Jiajun He, Xiyi Wang, Yaling Han, Xuchu Jiang, Chibiao Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13234125.txt.

Description

Hui Qi and colleagues released a 2026 study on computer diagnosis of brain tumors from MRI scans. Reading such scans by hand is slow, and nobody had compared the leading model families fairly under one strict protocol. Using 4,264 images from 333 patients, the team benchmarked convolutional networks and vision transformers with patient-level validation, so no person's scans leaked between training and testing.

A classic network, DenseNet-121, offered the best balance of accuracy and computing cost when no outside pretraining was allowed, while pretrained transformers stayed competitive when such data was available. The authors then upgraded DenseNet with coordinate attention, which tracks position along both image axes, and a hybrid image augmentation scheme. Their CAHA-Net topped the variants tested, reaching a macro F1 score near ninety-four percent with very few false alarms.

Catalog record

CREATORHui Qi, Ruizhe Guo, Jiajun He, Xiyi Wang, Yaling Han, Xuchu Jiang, Chibiao Liu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE80 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13234125

Browse the full library · About the Tower of Babel