Training Deep Networks for Prostate Cancer Diagnosis Using Coarse Histopathological Labels

Javadi, Golara; Samadi, Samareh; Bayat, Sharareh; Sojoudi, Samira; Hurtado, Antonio; Chang, Silvia D.; Black, Peter C.; Mousavi, Parvin; Abolmaesumi, Purang

doi:10.1007/978-3-030-87237-3_65

Cited by 12 publications

(7 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We hypothesize that the averaging from patch-wise to core-wise predictions may sufficiently smooth the effects of noisy labels at this level. We emphasize that the AUC for both methods is at least 10% higher than AUC achieved using conventional ultrasound machines [11], underlining the strong capabilities of high-frequency ultrasound.…”

Section: Effect Of Co-teachingmentioning

confidence: 77%

“…For each biopsy core X i , pathology reports a label Y i and the length of cancer L i in core, which is a rough estimate between zero and the biopsy sample length. Following previous work in PCa detection [14,11], we assign coarse pathology labels Y i to all extracted patches {x 1 , x 2 , ..., x ni } due to the lack of finer patchlevel labels. Therefore, many assigned labels to patches may not necessarily match with the ground truth and they are inherently weak.…”

Section: Methodsmentioning

confidence: 99%

“…We propose to use a state-of-the-art method, co-teaching, to address label noise for micro-ultrasound data [9]. For weak label methods, we rely on the findings of [11] showing the success of co-teaching method, and [20], which found that co-teaching significantly out-performed other methods such as robust loss functions. This approach simultaneously trains two similar neural networks with different weight initializations.…”

Section: Co-teachingmentioning

confidence: 99%

“…Machine learning models built from ultrasound data rely on ground truth labels from histopathology that are coarse and only approximately describe the spatial distribution of cancer in a biopsy core [11,14,16]. The lack of finer labels cause two challenges: first, labels assigned to patches of ultrasound images in a biopsy core may not match the ground truth tissue, resulting in weak labels; second, biopsies include other types of tissue such as fibromuscular cells, benign prostatic hyperplasia and precancerous changes.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Towards Confident Detection of Prostate Cancer using High Resolution Micro-ultrasound

Mahdi¹,

Wilson²,

Jamzad³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

MOTIVATION: Detection of prostate cancer during transrectal ultrasound-guided biopsy is challenging. The highly heterogeneous appearance of cancer, presence of ultrasound artefacts, and noise all contribute to these difficulties. Recent advancements in high-frequency ultrasound imaging -micro-ultrasound -have drastically increased the capability of tissue imaging at high resolution. Our aim is to investigate the development of a robust deep learning model specifically for micro-ultrasound-guided prostate cancer biopsy. For the model to be clinically adopted, a key challenge is to design a solution that can confidently identify the cancer, while learning from coarse histopathology measurements of biopsy samples that introduce weak labels. METH-ODS: We use a dataset of micro-ultrasound images acquired from 194 patients, who underwent prostate biopsy. We train a deep model using a co-teaching paradigm to handle noise in labels, together with an evidential deep learning method for uncertainty estimation. We evaluate the performance of our model using the clinically relevant metric of accuracy vs. confidence. RESULTS: Our model achieves a well-calibrated estimation of predictive uncertainty with area under the curve of 88%. The use of co-teaching and evidential deep learning in combination yields significantly better uncertainty estimation than either alone. We also provide a detailed comparison against state-of-the-art in uncertainty estimation.

show abstract

Section: Effect Of Co-teachingmentioning

confidence: 77%

Section: Methodsmentioning

confidence: 99%

Section: Co-teachingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Towards Confident Detection of Prostate Cancer using High Resolution Micro-ultrasound

Mahdi¹,

Wilson²,

Jamzad³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…For weak labeling, Zou et al [31] propose a noisy annotation tolerant network for breast ultrasound segmentation. Javadi et al propose to use multi-instance learning [32] and co-teaching [33] for PCa detection. Uncertainty estimation allows a model to express uncertainty when seeing OOD data rather than making false predictions; such methods have been applied to prostate [27], [34] and breast [35], [36] ultrasound to address heterogenity.…”

mentioning

confidence: 99%

Self-Supervised Learning with Limited Labeled Data for Prostate Cancer Detection in High Frequency Ultrasound

Wilson¹,

Mahdi²,

Jamzad³

et al. 2022

Preprint

View full text Add to dashboard Cite

Deep learning-based analysis of highfrequency, high-resolution micro-ultrasound data shows great promise for prostate cancer detection. Previous approaches to analysis of ultrasound data largely follow a supervised learning paradigm. Ground truth labels for ultrasound images used for training deep networks often include coarse annotations generated from the histopathological analysis of tissue samples obtained via biopsy. This creates inherent limitations on the availability and quality of labeled data, posing major challenges to the success of supervised learning methods. On the other hand, unlabeled prostate ultrasound data are more abundant. In this work, we successfully apply selfsupervised representation learning to micro-ultrasound data. Using ultrasound data from 1028 biopsy cores of 391 subjects obtained in two clinical centres, we demonstrate that feature representations learnt with this method can be used to classify cancer from non-cancer tissue, obtaining an AUROC score of 91% on an independent test set. To the best of our knowledge, this is the first successful end-to-end self-supervised learning approach for prostate cancer detection using ultrasound data. Our method outperforms baseline supervised learning approaches, generalizes well between different data centers, and scale well in performance as more unlabeled data are added, making it a promising approach for future research using large volumes of unlabeled data.

show abstract

Towards Confident Detection of Prostate Cancer Using High Resolution Micro-ultrasound

Mahdi

Wilson

Jamzad

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Training Deep Networks for Prostate Cancer Diagnosis Using Coarse Histopathological Labels

Cited by 12 publications

References 13 publications

Towards Confident Detection of Prostate Cancer using High Resolution Micro-ultrasound

Towards Confident Detection of Prostate Cancer using High Resolution Micro-ultrasound

Self-Supervised Learning with Limited Labeled Data for Prostate Cancer Detection in High Frequency Ultrasound

Towards Confident Detection of Prostate Cancer Using High Resolution Micro-ultrasound

Contact Info

Product

Resources

About