Division of Cancer Sciences, University of Manchester, Manchester, UK.
Radiotherapy Related Research, The Christie Foundation Trust, Manchester, UK.
Med Phys. 2022 May;49(5):3107-3120. doi: 10.1002/mp.15533. Epub 2022 Feb 28.
Skeletal muscle segmentation is an important procedure for assessing sarcopenia, an emerging imaging biomarker of patient frailty. Data annotation remains the bottleneck for training deep learning auto-segmentation models.
There is a need to define methodologies for applying models to different domains (e.g., anatomical regions or imaging modalities) without dramatically increasing data annotation.
To address this problem, we empirically evaluate the generalizability of various source tasks for transfer learning: natural image classification, natural image segmentation, unsupervised image reconstruction, and self-supervised jigsaw solving. Axial CT slices at L3 were extracted from PET-CT scans for 204 oesophago-gastric cancer patients and the skeletal muscle manually delineated by an expert. Features were transferred and segmentation models trained on subsets ( ) of the manually annotated training set. Four-fold cross-validation was performed to evaluate model generalizability. Human-level performance was established by performing an inter-observer study consisting of ten trained radiographers.
We find that accurate segmentation models can be trained on a fraction of the data required by current approaches. The Dice similarity coefficient and root mean square distance-to-agreement were calculated for each prediction and used to assess model performance. Models pre-trained on a segmentation task and fine-tuned on 10 images produce delineations that are comparable to those from trained observers and extract reliable measures of muscle health.
Appropriate transfer learning can generate convolutional neural networks for abdominal muscle segmentation that achieve human-level performance while decreasing the required data by an order of magnitude, compared to previous methods ( ). This work enables the development of future models for assessing skeletal muscle at other anatomical sites where large annotated data sets are scarce and clinical needs are yet to be addressed.
骨骼肌分割是评估肌肉减少症（一种新兴的患者脆弱性的影像学生物标志物）的重要程序。数据标注仍然是训练深度学习自动分割模型的瓶颈。
需要定义将模型应用于不同领域（例如解剖区域或成像方式）的方法，而无需大量增加数据标注。
为了解决这个问题，我们从经验上评估了各种源任务的可转移性，包括自然图像分类、自然图像分割、无监督图像重建和自监督拼图解决。从 204 名食管癌患者的 PET-CT 扫描中提取 L3 的轴向 CT 切片，并由专家手动勾画骨骼肌。在手动标注训练集的子集（ ）上进行特征转移和分割模型训练。采用四折交叉验证评估模型的泛化能力。通过进行由十位训练有素的放射技师组成的观察者间研究来建立人类水平的性能。
我们发现，可以在当前方法所需数据的一小部分上训练出准确的分割模型。为每个预测计算了 Dice 相似系数和均方根距离一致性，并用于评估模型性能。在分割任务上预训练并在 10 张图像上微调的模型产生的勾画与受过训练的观察者的勾画相当，并提取可靠的肌肉健康测量值。
适当的迁移学习可以生成用于腹部肌肉分割的卷积神经网络，与以前的方法相比，可将所需数据减少一个数量级，同时达到人类水平的性能（ ）。这项工作为在其他解剖部位评估骨骼肌的未来模型的开发提供了可能，这些部位的大型标注数据集稀缺，临床需求尚未得到解决。
