한빛사
>
한빛사논문
한빛사논문
- 조회678
Clin. Nucl. Med., Apr 16 2025, | https://doi.org/10.1097/RLU.0000000000005898
Comparison of CNNs and Transformer Models in Diagnosing Bone Metastases in Bone Scans Using Grad-CAM
Pak, Sehyun*; Son, Hye Joo MD, PhD†; Kim, Dongwoo MD‡; Woo, Ji Young MD, PhD§; Yang, Ik MD, PhD§; Hwang, Hee Sung MD, PhD‡; Rim, Dohyoung MS∥; Choi, Min Seok BS¶; Lee, Suk Hyun MD, PhD§
*Department of Medicine, Hallym University College of Medicine, Chuncheon, Gangwon, Republic of Korea
†Department of Nuclear Medicine, Dankook University Medical Center, Cheonan, Chungnam, Republic of Korea
‡Department of Nuclear Medicine, Hallym University Sacred Heart Hospital, Hallym University College of Medicine, Anyang, Gyeonggi, Republic of Korea
§Department of Radiology, Hallym University Kangnam Sacred Heart Hospital, Hallym University College of Medicine, Seoul, Republic of Korea
∥Rowan, Seoul, Republic of Korea
¶PE Data Solution, SK hynix, Icheon, Gyeonggi, Republic of Korea
S.P., H.J.S., and D.K. contributed equally as first authors.
Correspondence to: Suk Hyun Lee, MD, PhD
*Department of Medicine, Hallym University College of Medicine, Chuncheon, Gangwon, Republic of Korea
†Department of Nuclear Medicine, Dankook University Medical Center, Cheonan, Chungnam, Republic of Korea
‡Department of Nuclear Medicine, Hallym University Sacred Heart Hospital, Hallym University College of Medicine, Anyang, Gyeonggi, Republic of Korea
§Department of Radiology, Hallym University Kangnam Sacred Heart Hospital, Hallym University College of Medicine, Seoul, Republic of Korea
∥Rowan, Seoul, Republic of Korea
¶PE Data Solution, SK hynix, Icheon, Gyeonggi, Republic of Korea
S.P., H.J.S., and D.K. contributed equally as first authors.
Correspondence to: Suk Hyun Lee, MD, PhD
Abstract
Purpose: Convolutional neural networks (CNNs) have been studied for detecting bone metastases on bone scans; however, the application of ConvNeXt and transformer models has not yet been explored. This study aims to evaluate the performance of various deep learning models, including the ConvNeXt and transformer models, in diagnosing metastatic lesions from bone scans.
Materials and methods: We retrospectively analyzed bone scans from patients with cancer obtained at 2 institutions: the training and validation sets (n=4626) were from Hospital 1 and the test set (n=1428) was from Hospital 2. The deep learning models evaluated included ResNet18, the Data-Efficient Image Transformer (DeiT), the Vision Transformer (ViT Large 16), the Swin Transformer (Swin Base), and ConvNeXt Large. Gradient-weighted class activation mapping (Grad-CAM) was used for visualization.
Results: Both the validation set and the test set demonstrated that the ConvNeXt large model (0.969 and 0.885, respectively) exhibited the best performance, followed by the Swin Base model (0.965 and 0.840, respectively), both of which significantly outperformed ResNet (0.892 and 0.725, respectively). Subgroup analyses revealed that all the models demonstrated greater diagnostic accuracy for patients with polymetastasis compared with those with oligometastasis. Grad-CAM visualization revealed that the ConvNeXt Large model focused more on identifying local lesions, whereas the Swin Base model focused on global areas such as the axial skeleton and pelvis.
Conclusions: Compared with traditional CNN and transformer models, the ConvNeXt model demonstrated superior diagnostic performance in detecting bone metastases from bone scans, especially in cases of polymetastasis, suggesting its potential in medical image analysis.
Materials and methods: We retrospectively analyzed bone scans from patients with cancer obtained at 2 institutions: the training and validation sets (n=4626) were from Hospital 1 and the test set (n=1428) was from Hospital 2. The deep learning models evaluated included ResNet18, the Data-Efficient Image Transformer (DeiT), the Vision Transformer (ViT Large 16), the Swin Transformer (Swin Base), and ConvNeXt Large. Gradient-weighted class activation mapping (Grad-CAM) was used for visualization.
Results: Both the validation set and the test set demonstrated that the ConvNeXt large model (0.969 and 0.885, respectively) exhibited the best performance, followed by the Swin Base model (0.965 and 0.840, respectively), both of which significantly outperformed ResNet (0.892 and 0.725, respectively). Subgroup analyses revealed that all the models demonstrated greater diagnostic accuracy for patients with polymetastasis compared with those with oligometastasis. Grad-CAM visualization revealed that the ConvNeXt Large model focused more on identifying local lesions, whereas the Swin Base model focused on global areas such as the axial skeleton and pelvis.
Conclusions: Compared with traditional CNN and transformer models, the ConvNeXt model demonstrated superior diagnostic performance in detecting bone metastases from bone scans, especially in cases of polymetastasis, suggesting its potential in medical image analysis.
논문정보
- Research article
- 2025년 04월 (BRIC 등록일 2025-04-17)
- 국내 연구진
- 의약학 > 응용의학
관련 웨비나
![골스캔에서 골전이 진단에 Grad-CAM을 활용한 CNN과 트랜스포머 모델의 진단 성능 비교 [Clin. Nucl. Med.]](/upload/board/files/onseminar/thumb/thumb_0000940.png)
학술웨비나
골스캔에서 골전이 진단에 Grad-CAM을 활용한 CNN과 트랜스포머 모델의 진단 성능 비교 [Clin. Nucl. Med.] 목적: 골스캔(bone scan)에서 골 전이를 탐지하기 위해 합성곱 신경망(convolutional neural networks, CNNs)이 연구되어 왔으나, ConvNeXt와 트랜스포머(transformer) 모델은 아직 골스캔에 적용된 바가 없다. 본 연구는 ConvNeXt와 트랜스포머 모델을 포함한 다양한 딥러닝 모델이 골스캔에서 골 전이 병변을 진단하는 성능을 평가하는 것을 목표로 한다. 방법: 두 기관에서 얻은 암 환자 골스캔 영상을 후향적으로 분석하였다. 학습·검증 세트(n = 4,626)는 병원 1, 테스트 세트(n = 1,428)는 병원 2에서 수집하였다. 평가한 딥러닝 모델은 ResNet18, Data-Efficient Image Transformer(DeiT), Vision Transformer(ViT Large 16), Swin Transformer(Swin Base), ConvNeXt Large였다. 시각화를 위해 gradient-weighted class activation mapping(Grad-CAM)을 사용하였다. Results: ConvNeXt Large 모델은 검증 세트와 테스트 세트에서 각각 0.969와 0.885로 가장 우수한 성능을 보였으며, Swin Base 모델(0.965와 0.840)이 그 뒤를 이었고, 두 모델 모두 ResNet18(0.892와 0.725)보다 유의하게 우월하였다. 하위 분석에서 모든 모델은 소수 전이보다 다수 전이 환자에서 더 높은 진단 정확도를 나타냈다. Grad-CAM 시각화 결과 ConvNeXt Large 모델은 국소 병변에, Swin Base 모델은 몸통뼈대 및 골반과 같은 전역 영역에 더 집중하였다. Conclusions: ConvNeXt 모델은 기존 CNN 및 트랜스포머 모델보다 골스캔영상의 골 전이를 탐지하는 데 특히 다수 전이에서 뛰어난 진단 성능을 보여, 의료 영상 분석 분야에서의 활용 가능성을 시사한다.- 이석현(한림대학교)
댓글 작성 로그인
팝업 닫기관련 링크
소속기관 논문보기
관련분야 논문보기
