Impact of synthetic data on training a deep learning model for lesion detection and classification in contrast-enhanced mammography
Authors:
Open Access
- 28 April 2025
- journal article
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 12 (S2) , 1-18
Abstract
PurposePredictive models for contrast-enhanced mammography often perform better at detecting and classifying enhancing masses than (non-enhancing) microcalcification clusters. We aim to investigate whether incorporating synthetic data with simulated microcalcification clusters during training can enhance model performance.ApproachMicrocalcification clusters were simulated in low-energy images of lesion-free breasts from 782 patients, considering local texture features. Enhancement was simulated in the corresponding recombined images. A deep learning (DL) model for lesion detection and classification was trained with varying ratios of synthetic and real (850 patients) data. In addition, a handcrafted radiomics classifier was trained using delineations and class labels from real data, and predictions from both models were ensembled. Validation was performed on internal (212 patients) and external (279 patients) real datasets.ResultsThe DL model trained exclusively with synthetic data detected over 60% of malignant lesions. Adding synthetic data to smaller real training sets improved detection sensitivity for malignant lesions but decreased precision. Performance plateaued at a detection sensitivity of 0.80. The ensembled DL and radiomics models performed worse than the standalone DL model, decreasing the area under this receiver operating characteristic curve from 0.75 to 0.60 on the external validation set, likely due to falsely detected suspicious regions of interest.ConclusionsSynthetic data can enhance DL model performance, provided model setup and data distribution are optimized. The possibility to detect malignant lesions without real data present in the training set confirms the utility of synthetic data. It can serve as a helpful tool, especially when real data are scarce, and it is most effective when complementing real data.Keywords
Funding Information
- HORIZON EUROPE Framework Programme (952172)
- HORIZON EUROPE Framework Programme (952103)
- HORIZON EUROPE Framework Programme (101100633)
- HORIZON EUROPE Framework Programme (101095435)
- HORIZON EUROPE Framework Programme (101057699)
- HORIZON EUROPE Framework Programme (101034347)
- HORIZON EUROPE Framework Programme (101136670)
- Horizon 2020 Framework Programme (101057062)
This publication has 36 references indexed in Scilit:
- Calcifications at Digital Breast Tomosynthesis: Imaging Features and Biopsy TechniquesRadioGraphics, 2019
- Contrast-Enhanced Spectral Mammography in Women With Intermediate Breast Cancer Risk and Dense BreastsAmerican Journal of Roentgenology, 2018
- Radiomics: the bridge between medical imaging and personalized medicineNature Reviews Clinical Oncology, 2017
- Microcalcification on mammography: approaches to interpretation and biopsyBritish Journal of Radiology, 2017
- Computational growth model of breast microcalcification clusters in simulated mammographic environmentsComputers in Biology and Medicine, 2016
- Clinical utility of dual-energy contrast-enhanced spectral mammography for breast microcalcifications without associated mass: a preliminary analysisEuropean Radiology, 2015
- Low energy mammogram obtained in contrast-enhanced digital mammography (CEDM) is comparable to routine full-field digital mammography (FFDM)European Journal of Radiology, 2014
- Radiomics: Extracting more information from medical images using advanced feature analysisEuropean Journal of Cancer, 2012
- The simulation of 3D microcalcification clusters in 2D digital mammography and breast tomosynthesisMedical Physics, 2011
- Use of Microcalcification Descriptors in BI-RADS 4th Edition to Stratify Risk of MalignancyRadiology, 2007