Transfer Learning Through Weighted Loss Function and Group Normalization for Vessel Segmentation from Retinal Images

Sarhan, Abdullah; Rokne, Jon G.; Alhajj, Reda; Crichton, Andrew

doi:10.1109/icpr48806.2021.9412378

Cited by 11 publications

(13 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…"External'' means that validation data is from external datasets. The comparisons are with competitive methods of image quality grading 31 , binary vessel segmentation 39,57 , artery/vein segmentation 58 , and optic disc segmentation 59…”

Section: Resultsmentioning

confidence: 99%

“…In segmentation tasks, Intersection of Union (IoU) measures the overlap degree between ground truth maps and segmentation maps. Following the same setting 31,39,[57][58][59] , we set the ungradable images as the positive class in image quality grading. The probability of the ungradable category equals that of reject quality, and the probability of the gradable category is the sum of good quality and usable quality.…”

Section: Statistical Analyses and Compared Methodsmentioning

confidence: 99%

“…These methods used a reasonable proportion of data for model training and the remainder for internal validation (e.g., fivefold validation that means 80% of images are used for training and tuning, while 20% are used for validating the trained model) , and claimed that they have achieved state-of-the-art performance. As introduced in Table 1, the models of AutoMorph are trained on several public datasets and externally validated on separate datasets, while the compared methods 39,[57][58][59] are trained in the same domain data as the validation data, but with fewer training images. The goal of the comparison was not to prove the technical strengths of AutoMorph over recent methods, since this has already been verified in previously published work 22,23,47,48 .…”

Section: Statistical Analyses and Compared Methodsmentioning

confidence: 99%

“…All models are trained on the same training dataset and internally validated. As we don't have multiple groups of results for the compared methods 39,[57][58][59] , we cannot test if a statistically significant difference exists. But our results…”

Section: Confusion Matrix On Ddrmentioning

confidence: 99%

See 3 more Smart Citations

AutoMorph: Automated Retinal Vascular Morphology Quantification via a Deep Learning Pipeline

Zhou

Wagner

Chia

et al. 2022

Preprint

View full text Add to dashboard Cite

PurposeTo externally validate a deep learning pipeline (AutoMorph) for automated analysis of retinal vascular morphology on fundus photographs. AutoMorph has been made publicly available (https://github.com/rmaphoh/AutoMorph), facilitating widespread research in ophthalmic and systemic diseases.MethodsAutoMorph consists of four functional modules: image pre-processing, image quality grading, anatomical segmentation (including binary vessel, artery/vein, and optic disc/cup segmentation), and vascular morphology feature measurement. Image quality grading and anatomical segmentation use the most recent deep learning techniques. We employ a model ensemble strategy to achieve robust results and analyse the prediction confidence to rectify false gradable cases in image quality grading. We externally validate each module’s performance on several independent publicly available datasets.ResultsThe EfficientNet-b4 architecture used in the image grading module achieves comparable performance to the state-of-the-art for EyePACS-Q, with an F1-score of 0.86. The confidence analysis reduces the number of images incorrectly assessed as gradable by 76%. Binary vessel segmentation achieves an F1-score of 0.73 on AV-WIDE and 0.78 on DR-HAGIS. Artery/vein scores 0.66 on IOSTAR-AV, and disc segmentation achieves 0.94 in IDRID. Vascular morphology features measured from AutoMorph segmentation map and expert annotation show good to excellent agreement.ConclusionsAutoMorph modules perform well even when external validation data shows domain differences from training data, e.g., with different imaging devices. This fully automated pipeline can thus allow detailed, efficient and comprehensive analysis of retinal vascular morphology on colour fundus photographs.Translational RelevanceBy making AutoMorph publicly available and open source, we hope to facilitate ophthalmic and systemic disease research, particularly in the emerging field of ‘oculomics’.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Statistical Analyses and Compared Methodsmentioning

confidence: 99%

Section: Statistical Analyses and Compared Methodsmentioning

confidence: 99%

Section: Confusion Matrix On Ddrmentioning

confidence: 99%

See 2 more Smart Citations

AutoMorph: Automated Retinal Vascular Morphology Quantification via a Deep Learning Pipeline

Zhou

Wagner

Chia

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In segmentation tasks, IoU measures the overlap degree between ground-truth maps and segmentation maps. Following the same setting, 31 , 39 , 57 – 59 we set the ungradable images as the positive class in image quality grading. The probability of the ungradable category equals that of reject quality, and the probability of the gradable category is the sum of good quality and usable quality.…”

Section: Methodsmentioning

confidence: 99%

AutoMorph: Automated Retinal Vascular Morphology Quantification Via a Deep Learning Pipeline

Zhou

Wagner

Chia

et al. 2022

Trans. Vis. Sci. Tech.

View full text Add to dashboard Cite

Purpose To externally validate a deep learning pipeline (AutoMorph) for automated analysis of retinal vascular morphology on fundus photographs. AutoMorph has been made publicly available, facilitating widespread research in ophthalmic and systemic diseases. Methods AutoMorph consists of four functional modules: image preprocessing, image quality grading, anatomical segmentation (including binary vessel, artery/vein, and optic disc/cup segmentation), and vascular morphology feature measurement. Image quality grading and anatomical segmentation use the most recent deep learning techniques. We employ a model ensemble strategy to achieve robust results and analyze the prediction confidence to rectify false gradable cases in image quality grading. We externally validate the performance of each module on several independent publicly available datasets. Results The EfficientNet-b4 architecture used in the image grading module achieves performance comparable to that of the state of the art for EyePACS-Q, with an F 1 -score of 0.86. The confidence analysis reduces the number of images incorrectly assessed as gradable by 76%. Binary vessel segmentation achieves an F 1 -score of 0.73 on AV-WIDE and 0.78 on DR HAGIS. Artery/vein scores are 0.66 on IOSTAR-AV, and disc segmentation achieves 0.94 in IDRID. Vascular morphology features measured from the AutoMorph segmentation map and expert annotation show good to excellent agreement. Conclusions AutoMorph modules perform well even when external validation data show domain differences from training data (e.g., with different imaging devices). This fully automated pipeline can thus allow detailed, efficient, and comprehensive analysis of retinal vascular morphology on color fundus photographs. Translational Relevance By making AutoMorph publicly available and open source, we hope to facilitate ophthalmic and systemic disease research, particularly in the emerging field of oculomics.

show abstract

MR‐UNet: An UNet model using multi‐scale and residual convolutions for retinal vessel segmentation

Yang

Liu

2022

Int J Imaging Syst Tech

View full text Add to dashboard Cite

The overall performance of the retinal vessel segmentation network based on improved UNet is excellent, but there is still room for improvement in the small blood vessel segmentation. Therefore, this paper proposes an improved MR‐UNet, which designs two new blocks: the multi‐scale convolution (Multiconv) block and the residual convolution (Resconv) block. The Multiconv block uses different size convolution kernels to extract the characteristics of different thicknesses of retinal blood vessels, thereby improving the model's ability to segment small blood vessels. The Resconv block uses different convolutional layers to process the shallow semantic information in the encoding stage and then concatenates it to the decoding stage, reducing the semantic difference between the encoder and the decoder. On the retinal data sets DRIVE, STARE and CHASE_DB1, the accuracy (Acc) of this model is 0.9705, 0.9747, 0.9778, the specificity (Sp) is 0.9863, 0.9892, 0.9930, the AUC is 0.9872, 0.9849, 0.9925, and the F1‐Score is 0.8270, 0.8134, 0.8460, respectively. Compared with the original UNet, Se, Sp, and F1‐Score of MR‐UNet increase by 2.5%, 0.5%, and 0.4%, respectively, proving that MR‐UNet has better comprehensive performance.

show abstract

Transfer Learning Through Weighted Loss Function and Group Normalization for Vessel Segmentation from Retinal Images

Cited by 11 publications

References 29 publications

AutoMorph: Automated Retinal Vascular Morphology Quantification via a Deep Learning Pipeline

AutoMorph: Automated Retinal Vascular Morphology Quantification via a Deep Learning Pipeline

AutoMorph: Automated Retinal Vascular Morphology Quantification Via a Deep Learning Pipeline

MR‐UNet: An UNet model using multi‐scale and residual convolutions for retinal vessel segmentation

Contact Info

Product

Resources

About