Learning normalized inputs for iterative estimation in medical image segmentation

Drozdzal, Michal; Chartrand, Gabriel; Vorontsov, Eugene; Shakeri, Mahsa; Jorio, Lisa Di; Tang, An; Romero, Adriana; Bengio, Yoshua; Pal, Chris; Kadoury, Samuel

doi:10.1016/j.media.2017.11.005

Cited by 203 publications

(120 citation statements)

References 35 publications

Supporting

Mentioning

113

Contrasting

Order By: Relevance

“…It may be that if we had increased the capacity of the MIXED model by adding more convolutional layers to cope with the increased variability in the data, then we could have achieved more accurate results. Another possible approach suggested by Drozdxal et al . is to use a low capacity fully convolutional network as a preprocessor to normalize the MRI data first.…”

Section: Discussionmentioning

confidence: 99%

“…The input volume to the 3D-MIXED model is the FS volume. et al 32 is to use a low capacity fully convolutional network as a preprocessor to normalize the MRI data first. Alternatively, since we were able to achieve a DSC of 92% on just 14 training images, we could simply use domain adaptation and retrain a U-net using a small number of annotated volumes acquired using the desired sequence.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

An investigation of the effect of fat suppression and dimensionality on the accuracy of breast MRI segmentation using U‐nets

Fashandi

Kuling

et al. 2019

Medical Physics

View full text Add to dashboard Cite

Purpose Accurate segmentation of the breast is required for breast density estimation and the assessment of background parenchymal enhancement, both of which have been shown to be related to breast cancer risk. The MRI breast segmentation task is challenging, and recent work has demonstrated that convolutional neural networks perform well for this task. In this study, we have investigated the performance of several two‐dimensional (2D) U‐Net and three‐dimensional (3D) U‐Net configurations using both fat‐suppressed and nonfat‐suppressed images. We have also assessed the effect of changing the number and quality of the ground truth segmentations. Materials and methods We designed eight studies to investigate the effect of input types and the dimensionality of the U‐Net operations for the breast MRI segmentation. Our training data contained 70 whole breast volumes of T1‐weighted sequences without fat suppression (WOFS) and with fat suppression (FS). For each subject, we registered the WOFS and FS volumes together before manually segmenting the breast to generate ground truth. We compared four different input types to the U‐nets: WOFS, FS, MIXED (WOFS and FS images treated as separate samples), and MULTI (WOFS and FS images combined into a single multichannel image). We trained 2D U‐Nets and 3D U‐Nets with these data, which resulted in our eight studies (2D‐WOFS, 3D‐WOFS, 2D‐FS, 3D‐FS, 2D‐MIXED, 3D‐MIXED, 2D‐MULTI, and 3D‐MULT). For each of these studies, we performed a systematic grid search to tune the hyperparameters of the U‐Nets. A separate validation set with 15 whole breast volumes was used for hyperparameter tuning. We performed Kruskal–Walis test on the results of our hyperparameter tuning and did not find a statistically significant difference in the ten top models of each study. For this reason, we chose the best model as the model with the highest mean dice similarity coefficient (DSC) value on the validation set. The reported test results are the results of the top model of each study on our test set which contained 19 whole breast volumes annotated by three readers fused with the STAPLE algorithm. We also investigated the effect of the quality of the training annotations and the number of training samples for this task. Results The study with the highest average DSC result was 3D‐MULTI with 0.96 ± 0.02. The second highest average is 2D WOFS (0.96 ± 0.03), and the third is 2D MULTI (0.96 ± 0.03). We performed the Kruskal–Wallis one‐way ANOVA test with Dunn's multiple comparison tests using Bonferroni P‐value correction on the results of the selected model of each study and found that 3D‐MULTI, 2D‐MULTI, 3D‐WOFS, 2D‐WOFS, 2D‐FS, and 3D‐FS were not statistically different in their distributions, which indicates that comparable results could be obtained in fat‐suppressed and nonfat‐suppressed volumes and that there is no significant difference between the 3D and 2D approach. Our results also suggested that the networks trained on single sequence images or multiple sequence images organized in multichannel i...

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

An investigation of the effect of fat suppression and dimensionality on the accuracy of breast MRI segmentation using U‐nets

Fashandi

Kuling

et al. 2019

Medical Physics

View full text Add to dashboard Cite

show abstract

“…In the preprocessing step, both image intensity range normalization and histogram equalization methods were applied to reduce the computational time and improve the image contrast. 30 To normalize the image intensity range, all of the images were converted from 12-bit to 8-bit depths using both the maximum and minimum intensity values from the body region in question. As shown in Figure 6, whenever a high intensity region appeared in the soft tissue, the contrast decreased in the rectum region.…”

Section: Preprocessing Mr Imagementioning

confidence: 99%

Rectal cancer: Toward fully automatic discrimination of T2 and T3 rectal cancers using deep convolutional neural network

Kim

Lee

et al. 2019

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Preoperative chemoradiotherapy is known to reduce the local recurrence of locally advanced rectal cancer. However, the careful use of preoperative chemoradiotherapy is essential, because unnecessary over‐treatment can result in unintended complications. Therefore, a diagnostic system for distinguishing between T2 and T3 rectal cancers should be developed. According to the diagnostic criteria for rectal cancer, radiologists first identify the locations and the shapes of both the tumor and the rectum from a medical image and then diagnose the T2/T3 rectal cancer by determining whether the tumor passes through the rectal wall or not. We construct two distinct convolutional neural network models to achieve the automated segmentation of each rectum and tumor, respectively. Then, we construct another convolutional neural network model, which uses the output images of segmentation models as input and determines whether the tumor in the input magnetic resonance image is at the T2 stage or at the T3 stage. We evaluate the effectiveness of the proposed method based on 290 magnetic resonance images from 133 subjects. The proposed model demonstrates an accuracy of 94%.

show abstract

“…Less attention has been given to efficient algorithms performing simultaneous tracking of multiple interrelated objects [14] in order to eliminate the redundancies of tracking multiple objects via repeated use of single-object tracking. This problem is relevant to applications in medical imaging [10,11,22,29,30,38] as well as videos [12,43,55,58]. Figure 1.…”

Section: Introductionmentioning

confidence: 99%

Cross-Classification Clustering: An Efficient Multi-Object Tracking Technique for 3-D Instance Segmentation in Connectomics

Meirovitch

Saribekyan

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Pixel-accurate tracking of objects is a key element in many computer vision applications, often solved by iterated individual object tracking or instance segmentation followed by object matching. Here we introduce crossclassification clustering (3C), a technique that simultaneously tracks complex, interrelated objects in an image stack.The key idea in cross-classification is to efficiently turn a clustering problem into a classification problem by running a logarithmic number of independent classifications per image, letting the cross-labeling of these classifications uniquely classify each pixel to the object labels. We apply the 3C mechanism to achieve state-of-the-art accuracy in connectomics -the nanoscale mapping of neural tissue from electron microscopy volumes. Our reconstruction system increases scalability by an order of magnitude over existing single-object tracking methods (such as flood-filling networks). This scalability is important for the deployment of connectomics pipelines, since currently the best performing techniques require computing infrastructures that are beyond the reach of most laboratories. Our algorithm may offer benefits in other domains that require pixel-accurate tracking of multiple objects, such as segmentation of videos and medical imagery. * These authors equally contributed to this work

show abstract

Learning normalized inputs for iterative estimation in medical image segmentation

Cited by 203 publications

References 35 publications

An investigation of the effect of fat suppression and dimensionality on the accuracy of breast MRI segmentation using U‐nets

An investigation of the effect of fat suppression and dimensionality on the accuracy of breast MRI segmentation using U‐nets

Rectal cancer: Toward fully automatic discrimination of T2 and T3 rectal cancers using deep convolutional neural network

Cross-Classification Clustering: An Efficient Multi-Object Tracking Technique for 3-D Instance Segmentation in Connectomics

Contact Info

Product

Resources

About