Semi-supervised prostate cancer segmentation with multispectral MRI

Artan, Yusuf; Haider, Masoom A.; Langer, Deanne L.; Yetik, İmam Şamil

doi:10.1109/isbi.2010.5490091

Cited by 14 publications

(8 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Chung AG et al experimented with radiomics-drive CRF in 20 patients and achieved a DSC of only 0.391 (23). Other methods included: the Random Walker (RW) algorithm, where Artan Y et al achieved a DSC of 0.57±0.21 on 16 patients (24); the Relevance Vector Machine, where Ozer S et al achieved a DSC of 0.51 on 20 patients using the T2WI, ADC and 𝑘 𝑒𝑝 (25); and the Markov Random Field (MRF), where Liu X et al achieved a DSC of 0.6222 using T2, 𝑘 𝑙 , IAUC, ADC, T2W, A and 𝑘 𝑒𝑝 (14). However, all those methods were limited to the peripheral zone, and only a small number of patients were included in their studies.…”

Section: Discussionmentioning

confidence: 99%

Accurate Prostate Cancer Detection and Segmentation Using Non-Local Mask R-CNN With Histopathological Ground Truth

Dai

Jambor

Taimen

et al. 2021

International Journal of Radiation Oncology*Biology*Physics

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

Accurate Prostate Cancer Detection and Segmentation Using Non-Local Mask R-CNN With Histopathological Ground Truth

Dai

Jambor

Taimen

et al. 2021

International Journal of Radiation Oncology*Biology*Physics

View full text Add to dashboard Cite

show abstract

“…Intensity variations in MR data can significantly affect performances of many image processing techniques, hence, they need to be corrected (Madabhushi et al 2006a). Following the pre-processing procedure method described in Artan and Yetik (2012), Artan et al (2010), Liang and Bovik (2002), each image is median filtered to preserve edge boundaries. Subsequently, image intensities were normalised to zero mean unit variance and anisotropic diffusion filtering (Liang and Bovik 2002) is applied to remove noise.…”

Section: Pre-processingmentioning

confidence: 99%

“…Subsequently, image intensities were normalised to zero mean unit variance and anisotropic diffusion filtering (Liang and Bovik 2002) is applied to remove noise. This method is chosen because it does not cause inter regional blurring (Liang and Bovik 2002, Artan et al 2010, Artan and Yetik 2012. However, Artan and Yetik (2012) and Artan et al (2010) suggested that anisotropic diffusion filtering needs to be applied on the median-filtered and normalised images for better results.…”

Section: Pre-processingmentioning

confidence: 99%

“…This method is chosen because it does not cause inter regional blurring (Liang and Bovik 2002, Artan et al 2010, Artan and Yetik 2012. However, Artan and Yetik (2012) and Artan et al (2010) suggested that anisotropic diffusion filtering needs to be applied on the median-filtered and normalised images for better results. This three-step pre-processing approach has the following advantages (Artan et al 2010, Artan and Yetik 2012) (a) while suppressing the noise, it simultaneously preserves the edge boundaries (b) it standardise image intensities for all patients avoiding dissimilar intensity values for the same tissue types (c) it is a robust denoising method without blurring the tumor edges.…”

Section: Pre-processingmentioning

confidence: 99%

See 1 more Smart Citation

Computer-aided detection of prostate cancer in T2-weighted MRI within the peripheral zone

et al. 2016

View full text Add to dashboard Cite

In this paper we propose a prostate cancer computer-aided diagnosis (CAD) system and suggest a set of discriminant texture descriptors extracted from T2-weighted MRI data which can be used as a good basis for a multimodality system. For this purpose, 215 texture descriptors were extracted and eleven different classifiers were employed to achieve the best possible results. The proposed method was tested based on 418 T2-weighted MR images taken from 45 patients and evaluated using 9-fold cross validation with five patients in each fold. The results demonstrated comparable results to existing CAD systems using multimodality MRI. We achieved an area under the receiver operating curve (A z ) values equal to [Formula: see text], [Formula: see text], [Formula: see text] and [Formula: see text] for Bayesian networks, ADTree, random forest and multilayer perceptron classifiers, respectively, while a meta-voting classifier using average probability as a combination rule achieved [Formula: see text].

show abstract

“…Algorithms based on convolutional neural networks (CNNs) have shown promising results for PG segmentation of the whole PG and the PG zones (9, 10, 18-20, 11-15, 15-17). Even though CNNs perform well in PCa segmentation (16, [21][22][23][24], the training of the CNN remains challenging due to the absence of veri ed ground truth image data, as biopsy data is only available at a limited number of locations in the gland. Another problem of CNNs has been attributed to the intransparency associated with the way in which a CNN comes to a decision, which does not foster trust and acceptance amongst the end users.…”

Section: Introductionmentioning

confidence: 99%

Explainable AI for CNN-based Prostate Tumor Segmentation in Multi-parametric MRI Correlated to Whole Mount Histopathology

Gunashekar

Bielak

Hägele

et al. 2022

Preprint

View full text Add to dashboard Cite

Automatic prostate tumor segmentation is often unable to identify the lesion even if in multi-parametric MRI data is used as input, and the segmentation output is difficult to verify due to the lack of clinically established ground truth images. In this work we use an explainable deep learning model to interpret the predictions of a convolutional neural network (CNN) for prostate tumor segmentation. The CNN uses a U-Net architecture which was trained on multi-parametric MRI data from 122 patients to automatically segment the prostate gland and prostate tumor lesions. In addition, co-registered ground truth data from whole mount histopathology images were available in 15 patients that were used as a test set during CNN testing. To be able to interpret the segmentation results of the CNN, heat maps were generated using the Gradient Weighted Class Activation Map (Grad-CAM) method. With the CNN a mean Dice Sorensen Coefficient for the prostate gland and the tumor lesions of 0.62 and 0.31 with the radiologist drawn ground truth and 0.32 with wholemount histology ground truth for tumor lesions could be achieved. Dice Sorensen Coefficient between CNN predictions and manual segmentations from MRI and histology data were not significantly different. In the prostate the Grad-CAM heat maps could differentiate between tumor and healthy prostate tissue, which indicates that the image information in the tumor was essential for the CNN segmentation.

show abstract

Semi-supervised prostate cancer segmentation with multispectral MRI

Cited by 14 publications

References 13 publications

Accurate Prostate Cancer Detection and Segmentation Using Non-Local Mask R-CNN With Histopathological Ground Truth

Accurate Prostate Cancer Detection and Segmentation Using Non-Local Mask R-CNN With Histopathological Ground Truth

Computer-aided detection of prostate cancer in T2-weighted MRI within the peripheral zone

Explainable AI for CNN-based Prostate Tumor Segmentation in Multi-parametric MRI Correlated to Whole Mount Histopathology

Contact Info

Product

Resources

About