2026-10-10T11:24:26Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33815301 2026-09-15T18:06:04Z category_146 category_7 category_734 category_931 category_64 portal_5 item_type_3 month_year_09_2026

<p>Ablation study on the impact of key modules on model performance on the LIDC-IDRI dataset.</p> Lijun Zhang (483325) Debing Zhuo (24950643) Xiaolu Wu (6718964) Yuchou He (24950646) Guodong Kang (18459341) Space Science Medicine Biological Sciences not elsewhere classified Information Systems not elsewhere classified Cancer weak spatial priors three metrics among preserving contextual information displayed test cases concentrated mainly within complex nodule morphology based deep learning aware feature refinement related feature responses benign &# 8211 68 %, representing three view representations learned responses 448 benign 08 %, 06 %, weighted fusion vwf provided vwf ). view strategies view fusion view attention specific features sagittal slices relevant regions pulmonary nodules potential value models included malignant classification individual views highest values guided attention findings support dependent weights combined use baseline comparison adaptively aggregate ablation analysis 869 nodules 421 malignant <p>Ablation study on the impact of key modules on model performance on the LIDC-IDRI dataset.</p> 2026-09-15T18:05:45Z Dataset Dataset 10.1371/journal.pone.0357963.t003 https://figshare.com/articles/dataset/_p_Ablation_study_on_the_impact_of_key_modules_on_model_performance_on_the_LIDC-IDRI_dataset_p_/33815301 CC BY 4.0