Adaptive Contrast for Image Regression in Computer-Aided Disease Assessment

Dai, Weihang; Li, Xiaomeng; Chiu, K.W.; Kuo, Michael; Cheng, Kwang-Ting

doi:10.1109/tmi.2021.3137854

Cited by 17 publications

(10 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dai et al demonstrated a contrastive learning framework and evaluated bone mineral density estimation from X-ray images. 35 The images were fed into a feature extractor. Subsequently, a contrastive prediction branch and a regression branch were added after the feature layer for feature representation learning and regression prediction, respectively.…”

Section: Comparison Results Between the Proposed Sd-mdanet And Other ...mentioning

confidence: 99%

“…Dai et al. demonstrated a contrastive learning framework and evaluated bone mineral density estimation from X‐ray images 35 . The images were fed into a feature extractor.…”

Section: Resultsmentioning

confidence: 99%

“…Subsequently, a contrastive prediction branch and a regression branch were added after the feature layer for feature representation learning and regression prediction, respectively. To apply such method presented by Dai et al 35 . into this study, we replace input images with our CT images and the output of the regression branch with growth patterns arrays.…”

Section: Resultsmentioning

confidence: 99%

See 2 more Smart Citations

The self‐distillation trained multitask dense‐attention network for diagnosing lung cancers based on CT scans

Chen,

Zhang

2023

Medical Physics

View full text Add to dashboard Cite

BackgroundThe latest international multidisciplinary histopathological classification of lung cancer indicates that a deeper study of the lung adenocarcinoma requires a comprehensive multidisciplinary platform. However, in the traditional pathological examination or previous computer‐vision‐based research, the entire lung is not considered in a comprehensive manner.PurposeThe study aims to develop a deep learning model proposed for diagnosing the lung adenocarcinoma histopathologically based on CT scans. Instead of just classifying the lung adenocarcinoma, the pathological report should be inferred based on both the invasiveness and growth pattern of the tumors.MethodsA self‐distillation trained multitask dense‐attention network (SD‐MdaNet) is proposed and validated based on 2412 labeled CT scans from 476 patients and 845 unlabeled scans. Inferring the pathological report is divided into two tasks, predicting the invasiveness of the lung tumor and inferring growth patterns of tumor cells in a comprehensive histopathological subtyping manner with excellent accuracy. In the proposed method, the dense‐attention module is introduced to better extract features from a small dataset in the main branch of the MdaNet. Next, task‐specific attention modules are utilized in different branches and finally integrated as a multitask model. The second task is a blend of classification and regression tasks. Thus, a specialized loss function is developed. In the proposed knowledge distillation (KD) process, the MdaNet as well as its main branch trained for solving two single tasks, respectively, are treated as multiple teachers to produce a student model. A novel KD loss function is developed to take the advantage of all the models as well as data with labels and without labels.ResultsSD‐MdaNet achieves an AUC of on invasiveness prediction, and on predominant growth pattern prediction on our dataset. Moreover, the average mean squared error in inferring growth pattern proportion reaches 0.0217 ± 0.0019, and the AUC for predominant growth pattern proportion reaches . The proposed SD‐MdaNet is significantly better than all other benchmarking methods ().ConclusionsExperimental results demonstrate that the proposed SD‐MdaNet can significantly improve the performance of the lung adenocarcinoma pathological diagnosis using only CT scans. Analyses and discussions are conducted to interpret the advantages of the SD‐MdaNet.

show abstract

Section: Comparison Results Between the Proposed Sd-mdanet And Other ...mentioning

confidence: 99%

“…Dai et al. demonstrated a contrastive learning framework and evaluated bone mineral density estimation from X‐ray images 35 . The images were fed into a feature extractor.…”

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

The self‐distillation trained multitask dense‐attention network for diagnosing lung cancers based on CT scans

Chen,

Zhang

2023

Medical Physics

View full text Add to dashboard Cite

show abstract

“…The main purpose of deep metric learning and contrastive learning is to train a deep learning model to learn the distance feature representations in an embedding space. The main difference between these two methods is that contrastive learning is closely related to the self-supervised learning domain, which contains a data augmentation step to generalize an arbitrary number of positive and negative sample pairs from each sample [32]. Given the stunning achievement of selfsupervised representation learning, many contrastive learning methods have been developed for various computer vision tasks [33]- [36].…”

Section: B Contrastive Representation Learningmentioning

confidence: 99%

Learning Generalized Hybrid Proximity Representation for Image Recognition

Li¹,

Ralescu²

2023

Preprint

View full text Add to dashboard Cite

Recently, deep metric learning techniques received attentions, as the learned distance representations are useful to capture the similarity relationship among samples and further improve the performance of various of supervised or unsupervised learning tasks. We propose a novel supervised metric learning method that can learn the distance metrics in both geometric and probabilistic space for image recognition. In contrast to the previous metric learning methods which usually focus on learning the distance metrics in Euclidean space, our proposed method is able to learn better distance representation in a hybrid approach. To achieve this, we proposed a Generalized Hybrid Metric Loss (GHM-Loss) to learn the general hybrid proximity features from the image data by controlling the tradeoff between geometric proximity and probabilistic proximity. To evaluate the effectiveness of our method, we first provide theoretical derivations and proofs of the proposed loss function, then we perform extensive experiments on two public datasets to show the advantage of our method compared to other stateof-the-art metric learning methods.

show abstract

“…However, regression tasks require continuous labels (e.g., neurocognitive scores) that cannot directly be used for pair determination. Two recent works have shown that contrastive learning can be useful in the context of regression based on medical images as input (Lei et al, 2021;Dai et al, 2022). For example, RPR-Loc proposed a learning strategy to predict the distance between a pair of image patches (Lei et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

Data_Sheet_1.PDF

View full text Add to dashboard Cite

Adaptive Contrast for Image Regression in Computer-Aided Disease Assessment

Cited by 17 publications

References 52 publications

The self‐distillation trained multitask dense‐attention network for diagnosing lung cancers based on CT scans

The self‐distillation trained multitask dense‐attention network for diagnosing lung cancers based on CT scans

Learning Generalized Hybrid Proximity Representation for Image Recognition

Data_Sheet_1.PDF

Contact Info

Product

Resources

About