What does this research mean for the field?

A habitat-guided 2.5D multiparametric MRI deep learning model can noninvasively and accurately predict Ki-67 expression status in breast cancer, outperforming conventional 2D deep learning and clinical models. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

Develop and validate a habitat-guided 2.5D deep learning model based on multiparametric MRI for noninvasive preoperative prediction of Ki-67 expression in invasive breast cancer.

September 3, 2026Journal of Computer Assisted Tomography

Prediction of Ki-67 Status in Breast Cancer Using a Habitat-Guided 2.5D Multiparametric MRI Deep Learning Model

View Full Paper

Ask AI

Bookmark

Authors

ZMZeyang MiaoShanghai University of Traditional Chinese Medicine RXRun Xu MGMengyao GuoLudong University

Discussion

Member takes

Overview

Retrospective study demonstrates noninvasive Ki-67 status prediction in breast cancer using habitat-guided deep learning MRI, suggesting potential for noninvasive preoperative risk assessment.

Key Points

Develop and validate a habitat-guided 2.5D deep learning model based on multiparametric MRI for noninvasive preoperative prediction of Ki-67 expression in invasive breast cancer.
Retrospective multicohort study involving 333 patients with invasive breast carcinoma from two distinct MRI vendor cohorts (training set, n=233; independent test set, n=100).
Dynamic contrast-enhanced MRI and diffusion-weighted imaging parameters were clustered via K-means (k=3) into functional habitat masks to create a 7-channel 2.5D input tensor evaluated on a ResNet18 backbone.
In the independent cross-vendor test set, the habitat-guided 2.5D model achieved an AUC of 0.821 (95% CI: 0.736–0.906) and a sensitivity of 0.804.
The habitat-guided model significantly outperformed a conventional 2D deep learning model (AUC: 0.654, P=0.002) and a clinical model (AUC: 0.686, P=0.019), with clinical variables providing no incremental benefit (combined AUC: 0.837, P=0.483).

Cite This Study

Miao et al. (2026) studied this question.

synapsesocial.com/papers/6a99351f636c6408cfa7d0f0 https://doi.org/10.1097/rct.0000000000001923

View Full Paper

Ask AI

Bookmark

September 3, 2026Journal of Computer Assisted Tomography

Prediction of Ki-67 Status in Breast Cancer Using a Habitat-Guided 2.5D Multiparametric MRI Deep Learning Model

View Full Paper

Ask AI

Bookmark

Authors

ZMZeyang MiaoShanghai University of Traditional Chinese Medicine RXRun Xu MGMengyao GuoLudong University

Discussion

Member takes

Overview

Retrospective study demonstrates noninvasive Ki-67 status prediction in breast cancer using habitat-guided deep learning MRI, suggesting potential for noninvasive preoperative risk assessment.

Key Points

Develop and validate a habitat-guided 2.5D deep learning model based on multiparametric MRI for noninvasive preoperative prediction of Ki-67 expression in invasive breast cancer.
Retrospective multicohort study involving 333 patients with invasive breast carcinoma from two distinct MRI vendor cohorts (training set, n=233; independent test set, n=100).
Dynamic contrast-enhanced MRI and diffusion-weighted imaging parameters were clustered via K-means (k=3) into functional habitat masks to create a 7-channel 2.5D input tensor evaluated on a ResNet18 backbone.
In the independent cross-vendor test set, the habitat-guided 2.5D model achieved an AUC of 0.821 (95% CI: 0.736–0.906) and a sensitivity of 0.804.
The habitat-guided model significantly outperformed a conventional 2D deep learning model (AUC: 0.654, P=0.002) and a clinical model (AUC: 0.686, P=0.019), with clinical variables providing no incremental benefit (combined AUC: 0.837, P=0.483).

Cite This Study

Miao et al. (2026) studied this question.

synapsesocial.com/papers/6a99351f636c6408cfa7d0f0 https://doi.org/10.1097/rct.0000000000001923

View Full Paper

Ask AI

Bookmark