TY - JOUR
T1 - Combining Deep Learning and Handcrafted Radiomics for Classification of Suspicious Lesions on Contrast-enhanced Mammograms
AU - Beuque, Manon P.L.
AU - Lobbes, Marc B.I.
AU - van Wijk, Yvonka
AU - Widaatalla, Yousif
AU - Primakov, Sergey
AU - Majer, Michael
AU - Balleyguier, Corinne
AU - Woodruff, Henry C.
AU - Lambin, Philippe
N1 - Funding Information:
The authors acknowledge financial support from ERC advanced grant (ERC-ADG-2015 no. 694812 – Hypoximmuno) and ERC-2020-PoC: 957565-AUTO.DISTINCT. The authors also acknowledge financial support from EUROSTARS (DEEP-MAM: ESTAR19103), the European Union’s Horizon research and innovation programme under grant agreement MSCA-ITN-PREDICT no. 766276, CHAIMELEON no. 952172, EuCanImage no. 952103, TRANSCAN Joint Transnational Call 2016 (JTC2016 CLEARLY no. UM 2017-8295), IMI-OPTIMA no. 101034347, AIDAVA (HORIZON-HLTH-2021-TOOL-06) no.101057062, and EUCAIM (DIGITAL-2022-CLOUD-AI-02) no.101100633. * H.C.W. and P.L. are co–senior authors. Conflicts of interest are listed at the end of this article. See also the editorial by Bahl and Do in this issue.
Funding Information:
Research grant from GE Healthcare; speaker’s fees from GE Healthcare, Tromp Medical, Bayer, and Guerbet; medical advisory board member for Bayer, Hologic, and GE Healthcare. Y.v.W. No relevant relationships. Y.W. No relevant relationships. S.P. Grant from Predict Innovative Training Network. M.M. No relevant relationships. C.B. No relevant relationships. H.C.W. Minority shares in Radiomics. P.L. Consulting fees from Radiomics and Benelux Health Ventures; co-inventor on patents licensed to Radiomics and has licensed software to Radiomics; unpaid advisory board member for Radiomics, Communicare Biotech, LivingMed Biotech, Bactam, and Convert Pharmaceuticals; minority shares in Radiomics and Com-municare Solutions.
Publisher Copyright:
© RSNA, 2023.
PY - 2023/6/1
Y1 - 2023/6/1
N2 - Background Handcrafted radiomics and deep learning (DL) models individually achieve good performance in lesion classification (benign vs malignant) on contrast-enhanced mammography (CEM) images. Purpose To develop a comprehensive machine learning tool able to fully automatically identify, segment, and classify breast lesions on the basis of CEM images in recall patients. Materials and Methods CEM images and clinical data were retrospectively collected between 2013 and 2018 for 1601 recall patients at Maastricht UMC+ and 283 patients at Gustave Roussy Institute for external validation. Lesions with a known status (malignant or benign) were delineated by a research assistant overseen by an expert breast radiologist. Preprocessed low-energy and recombined images were used to train a DL model for automatic lesion identification, segmentation, and classification. A handcrafted radiomics model was also trained to classify both human- and DL-segmented lesions. Sensitivity for identification and the area under the receiver operating characteristic curve (AUC) for classification were compared between individual and combined models at the image and patient levels. Results After the exclusion of patients without suspicious lesions, the total number of patients included in the training, test, and validation data sets were 850 (mean age, 63 years ± 8 [SD]), 212 (62 years ± 8), and 279 (55 years ± 12), respectively. In the external data set, lesion identification sensitivity was 90% and 99% at the image and patient level, respectively, and the mean Dice coefficient was 0.71 and 0.80 at the image and patient level, respectively. Using manual segmentations, the combined DL and handcrafted radiomics classification model achieved the highest AUC (0.88 [95% CI: 0.86, 0.91]) (P < .05 except compared with DL, handcrafted radiomics, and clinical features model, where P = .90). Using DL-generated segmentations, the combined DL and handcrafted radiomics model showed the highest AUC (0.95 [95% CI: 0.94, 0.96]) (P < .05). Conclusion The DL model accurately identified and delineated suspicious lesions on CEM images, and the combined output of the DL and handcrafted radiomics models achieved good diagnostic performance. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Bahl and Do in this issue.
AB - Background Handcrafted radiomics and deep learning (DL) models individually achieve good performance in lesion classification (benign vs malignant) on contrast-enhanced mammography (CEM) images. Purpose To develop a comprehensive machine learning tool able to fully automatically identify, segment, and classify breast lesions on the basis of CEM images in recall patients. Materials and Methods CEM images and clinical data were retrospectively collected between 2013 and 2018 for 1601 recall patients at Maastricht UMC+ and 283 patients at Gustave Roussy Institute for external validation. Lesions with a known status (malignant or benign) were delineated by a research assistant overseen by an expert breast radiologist. Preprocessed low-energy and recombined images were used to train a DL model for automatic lesion identification, segmentation, and classification. A handcrafted radiomics model was also trained to classify both human- and DL-segmented lesions. Sensitivity for identification and the area under the receiver operating characteristic curve (AUC) for classification were compared between individual and combined models at the image and patient levels. Results After the exclusion of patients without suspicious lesions, the total number of patients included in the training, test, and validation data sets were 850 (mean age, 63 years ± 8 [SD]), 212 (62 years ± 8), and 279 (55 years ± 12), respectively. In the external data set, lesion identification sensitivity was 90% and 99% at the image and patient level, respectively, and the mean Dice coefficient was 0.71 and 0.80 at the image and patient level, respectively. Using manual segmentations, the combined DL and handcrafted radiomics classification model achieved the highest AUC (0.88 [95% CI: 0.86, 0.91]) (P < .05 except compared with DL, handcrafted radiomics, and clinical features model, where P = .90). Using DL-generated segmentations, the combined DL and handcrafted radiomics model showed the highest AUC (0.95 [95% CI: 0.94, 0.96]) (P < .05). Conclusion The DL model accurately identified and delineated suspicious lesions on CEM images, and the combined output of the DL and handcrafted radiomics models achieved good diagnostic performance. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Bahl and Do in this issue.
U2 - 10.1148/radiol.221843
DO - 10.1148/radiol.221843
M3 - Article
C2 - 37338353
SN - 0033-8419
VL - 307
SP - 221843
JO - Radiology
JF - Radiology
IS - 5
M1 - e221843
ER -