Habitat-Radiomics combining multichannel 2.5D deep learning for differentiating adrenal adenomas from metastases using automatic segmentation: a multicenter study - Takeaways - MDSpire

Habitat-Radiomics combining multichannel 2.5D deep learning for differentiating adrenal adenomas from metastases using automatic segmentation: a multicenter study

  • By

  • shengnan, Yin

  • Ning, Ding

  • Chuqi, Yang

  • Shaocai, Wang

  • Mengjuan, Li

  • Ji, Yiding

  • Tong, Liu

  • Long, Jin

  • May 27, 2026

  • 0 min

Share

  • 1

    The study involved 390 patients from two hospitals, divided into training, internal validation, and external test sets.

  • 2

    Lesion segmentation was performed using the Medical SAM model, and a 2.5D deep learning model was built with DenseNet-121 architecture.

  • 3

    Habitat-radiomics utilized K-means clustering, with features extracted using PyRadiomics and XGBoost models developed.

  • 4

    The fusion model achieved AUCs of 0.983, 0.913, and 0.886 in training, internal validation, and external test sets, respectively.

  • 5

    The fusion model may aid in noninvasively differentiating lipid-poor adrenal adenomas from metastases for precision treatment.

Original Source(s)

Habitat-Radiomics combining multichannel 2.5D deep learning for differentiating adrenal adenomas from metastases using automatic segmentation: a multicenter study

  • by shengnan, Yin , Ning, Ding , Chuqi, Yang , Shaocai, Wang , Mengjuan, Li , Ji, Yiding , Tong, Liu , Long, Jin

  • May 27, 2026

Related Content

POU2F3-positive small cell neuroendocrine carcinoma of the urinary tract: a clinicopathological analysis of five cases

Viable CAR T-cells remain detectable in cerebrospinal fluid in patients with grade ≥3 ICANS despite corticosteroid therapy

Editorial: Importance of quality assurance in radiation oncology clinical trials