Prostate Cancer Detection using Deep Convolutional Neural Networks

Yoo, Sunghwan; Gujrathi, Isha; Haider, Masoom A.; Khalvati, Farzad

doi:10.1038/s41598-019-55972-4

Cited by 174 publications

(123 citation statements)

References 43 publications

Supporting

Mentioning

119

Contrasting

Unclassified

Order By: Relevance

“…The histopathological standard of reference was MR-TRUS fusion biopsy and not radical prostatectomy (RP). However, the sensitivity of the extended systematic and targeted biopsy performed here has previously been shown to detect 97 % of sPC compared to RP specimen [12]. In addition, only a cohort based on biopsies can encompass all men that are important to consider in a screening setting of men with suspicion for sPC.…”

Section: Discussionmentioning

confidence: 90%

“…Novel artificial intelligence approaches such as convolutional neural networks (CNN) promise to capture diagnostically decisive information directly from medical images [9,10]. In the prostate, systems providing fully automated prostate assessment and lesion segmentation [11] or based on slice classification [12] have been developed. Other applications have utilized CNNs to evaluate predefined regions on prostate MR images [13].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Comparison of Prostate MRI Lesion Segmentation Agreement Between Multiple Radiologists and a Fully Automatic Deep Learning System

Schelb

Tavakoli

Tubtawee

et al. 2020

Rofo

View full text Add to dashboard Cite

Purpose A recently developed deep learning model (U-Net) approximated the clinical performance of radiologists in the prediction of clinically significant prostate cancer (sPC) from prostate MRI. Here, we compare the agreement between lesion segmentations by U-Net with manual lesion segmentations performed by different radiologists. Materials and Methods 165 patients with suspicion for sPC underwent targeted and systematic fusion biopsy following 3 Tesla multiparametric MRI (mpMRI). Five sets of segmentations were generated retrospectively: segmentations of clinical lesions, independent segmentations by three radiologists, and fully automated bi-parametric U-Net segmentations. Per-lesion agreement was calculated for each rater by averaging Dice coefficients with all overlapping lesions from other raters. Agreement was compared using descriptive statistics and linear mixed models. Results The mean Dice coefficient for manual segmentations showed only moderate agreement at 0.48–0.52, reflecting the difficult visual task of determining the outline of otherwise jointly detected lesions. U-net segmentations were significantly smaller than manual segmentations (p < 0.0001) and exhibited a lower mean Dice coefficient of 0.22, which was significantly lower compared to manual segmentations (all p < 0.0001). These differences remained after correction for lesion size and were unaffected between sPC and non-sPC lesions and between peripheral and transition zone lesions. Conclusion Knowledge of the order of agreement of manual segmentations of different radiologists is important to set the expectation value for artificial intelligence (AI) systems in the task of prostate MRI lesion segmentation. Perfect agreement (Dice coefficient of one) should not be expected for AI. Lower Dice coefficients of U-Net compared to manual segmentations are only partially explained by smaller segmentation sizes and may result from a focus on the lesion core and a small relative lesion center shift. Although it is primarily important that AI detects sPC correctly, the Dice coefficient for overlapping lesions from multiple raters can be used as a secondary measure for segmentation quality in future studies. Key Points: Citation Format

show abstract

Section: Discussionmentioning

confidence: 90%

Section: Introductionmentioning

confidence: 99%

Comparison of Prostate MRI Lesion Segmentation Agreement Between Multiple Radiologists and a Fully Automatic Deep Learning System

Schelb

Tavakoli

Tubtawee

et al. 2020

Rofo

View full text Add to dashboard Cite

show abstract

“…Prostate cancer (PCa) is the most common form of cancer among males. But if detected in the early stages, because of the slow progression of the disease, the survival rates are high (Yoo et al 2019 ; Chen et al 2019 ) . The problem is typically observed in men with middle age or older (Yoon et al 2016 ).…”

Section: Introductionmentioning

confidence: 99%

HPLC phenolic profile and induction of apoptosis by Linum usitatissimum extract in LNCaP cells by caspase3 and Bax pathways

et al. 2020

View full text Add to dashboard Cite

Linum usitatissimum is a candidate as a remedy to treat prostate problems in some folklore medicines. In this study, we have reported the phenolic and flavonoid constituents, antioxidant activity, and potential of the plant extract against prostate cancer cells. The phenolic and flavonoid compound profile of the extract were established using HPLC analysis. While the total phenolic and flavonoid content (TPC and TFC) were analyzed using classic methods. The antioxidant activity of the extract was also evaluated. MTT assay and flow cytometry technique was used to evaluate antiproliferation activity and induction apoptosis of the plant extract on prostate cancer cells of LNCaP. We also evaluated the gene expression of Bax and caspase-3 using the real-time qPCR assay. HPLC result revealed that L. usitatissimum extract (LUE) was rich in phenolic acids such as gallic, ferulic, and vanillic acid with the amount of 3.56, 2.12, 1.24 μg/g extract respectively. 383.4 mg GAE/g and 47.1 mgRuE/g were calculated for total phenolic and flavonoid content. LUE exhibited radical scavenging activity with IC50 = 19.3 ± 1.1 µg/mL. LUE chelated ferrous ions with IC50 = 121.1 ± 1.3 µg/mL. LUE showed anti-proliferative activity on LNCaP cells with the IC50 values of 8.3, 6.3, and 5.4 μg/mL after 24, 48, and 72 h treatment. LUE also increased cell mortality by inducing apoptosis (15.3–29.8%). The real-time qPCR results exhibited an increase in gene expression of Bax and caspase-3. Our in vitro study demonstrates that L. usitatissimum can be considered as an effective agent to inhibit the growth and invasion the human prostate cancer cells.

show abstract

“…Then, cropped CT scans are placed in black images, whereas cropped PET scans are placed in black images. Determining the size of the inputs, based on the largest tumor, to ensure all the target area is covered, is a standard practice in deep learning-based cancer image analysis 57 . As the inputs to our model are 3D images, where the third dimension is of size 3, three cropped slices, for each tumor, are stacked together.…”

Section: Methodsmentioning

confidence: 99%

$$\text {DRTOP}$$: deep learning-based radiomics for the time-to-event outcome prediction in lung cancer

Afshar

Mohammadi

Tyrrell

et al. 2020

Sci Rep

View full text Add to dashboard Cite

Hand-crafted radiomics has been used for developing models in order to predict time-to-event clinical outcomes in patients with lung cancer. Hand-crafted features, however, are pre-defined and extracted without taking the desired target into account. furthermore, accurate segmentation of the tumor is required for development of a reliable predictive model, which may be objective and a time-consuming task. to address these drawbacks, we propose a deep learning-based radiomics model for the time-to-event outcome prediction, referred to as DRtop that takes raw images as inputs, and calculates the image-based risk of death or recurrence, for each patient. our experiments on an in-house dataset of 132 lung cancer patients show that the obtained image-based risks are significant predictors of the time-to-event outcomes. Computed Tomography (CT)-based features are predictors of the overall survival (OS), with the hazard ratio (HR) of 1.35, distant control (DC), with HR of 1.06, and local control (LC), with HR of 2.66. The Positron Emission Tomography (PET)-based features are predictors of OS and recurrence free survival (RFS), with hazard ratios of 1.67 and 1.18, respectively. the concordance indices of 68% , 63% , and 64% for predicting the oS, Dc, and RfS show that the deep learning-based radiomics model is as accurate or better in predicting predefined clinical outcomes compared to hand-crafted radiomics, with concordance indices of 51% , 64% , and 47% , for predicting the OS, DC, and RFS, respectively. Deep learning-based radiomics has the potential to offer complimentary predictive information in the personalized management of lung cancer patients. Despite significant advancements in treatment, lung cancer remains the leading cause of cancer-related mortalities worldwide 1. Lung cancer is among the most common cancers and, together with breast cancer, includes most of the newly diagnosed cancer cases 2. Significant recent progress in the biological understanding and tumor heterogeneity of non-small cell lung cancer calls for treatment individualization. Specific clinical endpoints are used in clinical trials to measure the clinical benefit of a specific treatment 3,4. Although overall survival (OS) remains the gold standard, other clinical endpoints such as recurrence free survival (RFS), distant control (DC), and local control (LC) measure different and significant aspects of the clinical benefit of treatment. Inherent difficulties to assess these clinical outcomes such as the lengthy duration of the follow-up needed until the time of event and the various parameters, unrelated to the primary cancer, affecting the result during follow-up, have led to a surge for developing surrogates that can predict clinical outcomes noninvasively. Recently, radiomics, which is the process of extracting high throughput quantitative and semi-quantitative features from medical images aiming at diagnosis, classification or prediction of outcomes, has attracted much attention, showing promising results 5-15. Studies, investigating th...

show abstract

Prostate Cancer Detection using Deep Convolutional Neural Networks

Cited by 174 publications

References 43 publications

Comparison of Prostate MRI Lesion Segmentation Agreement Between Multiple Radiologists and a Fully Automatic Deep Learning System

Comparison of Prostate MRI Lesion Segmentation Agreement Between Multiple Radiologists and a Fully Automatic Deep Learning System

HPLC phenolic profile and induction of apoptosis by Linum usitatissimum extract in LNCaP cells by caspase3 and Bax pathways

$$\text {DRTOP}$$: deep learning-based radiomics for the time-to-event outcome prediction in lung cancer

Contact Info

Product

Resources

About