MRI-based prostate and dominant lesion segmentation using cascaded scoring convolutional neural network

PURPOSE

Dose escalation to dominant intraprostatic lesions (DILs) is a novel treatment strategy to improve the treatment outcome of prostate radiation therapy. Treatment planning requires accurate and fast delineation of the prostate and DILs. In this study, a 3D cascaded scoring convolutional neural network is proposed to automatically segment the prostate and DILs from MRI.

METHODS AND MATERIALS

The proposed cascaded scoring convolutional neural network performs end-to-end segmentation by locating a region-of-interest (ROI), identifying the object within the ROI, and defining the target. A scoring strategy, which is learned to judge the segmentation quality of DIL, is integrated into cascaded convolutional neural network to solve the challenge of segmenting the irregular shapes of the DIL. To evaluate the proposed method, 77 patients who underwent MRI and PET/CT were retrospectively investigated. The prostate and DIL ground truth contours were delineated by experienced radiologists. The proposed method was evaluated with five-fold cross validation and holdout testing.

RESULTS

The average centroid distance, volume difference, and Dice similarity coefficient (DSC) value for prostate/DIL are 4.3±7.5mm/3.73±3.78mm, 4.5±7.9cc/0.41±0.59cc and 89.6±8.9%/84.3±11.9%, respectively. Comparable results were obtained in the holdout test. Similar or superior segmentation outcomes were seen when compared the results of the proposed method to those of competing segmentation approaches CONCLUSIONS: : The proposed automatic segmentation method can accurately and simultaneously segment both the prostate and DILs. The intended future use for this algorithm is focal boost prostate radiation therapy. This article is protected by copyright. All rights reserved.

Prostate and dominant intraprostatic lesion segmentation on PET/CT using cascaded regional-net

Focal boost to dominant intraprostatic lesions (DILs) has recently been proposed for prostate radiation therapy. Accurate and fast delineation of the prostate and DILs is thus required during treatment planning. In this paper, we develop a learning-based method using positron emission tomography (PET)/computed tomography (CT) images to automatically segment the prostate and its DILs. To enable end-to-end segmentation, a deep learning-based method, called cascaded regional-Net, is utilized. The first network, referred to as dual attention network, is used to segment the prostate via extracting comprehensive features from both PET and CT images. A second network, referred to as mask scoring regional convolutional neural network (MSR-CNN), is used to segment the DILs from the PET and CT within the prostate region. Scoring strategy is used to diminish the misclassification of the DILs. For DIL segmentation, the proposed cascaded regional-Net uses two steps to remove normal tissue regions, with the first step cropping images based on prostate segmentation and the second step using MSR-CNN to further locate the DILs. The binary masks of DILs and prostates of testing patients are generated on the PET/CT images by the trained model. For evaluation, we retrospectively investigated 49 prostate cancer patients with PET/CT images acquired. The prostate and DILs of each patient were contoured by radiation oncologists and set as the ground truths and targets. We used five-fold cross-validation and a hold-out test to train and evaluate our method. The mean surface distance and DSC values were 0.666 ± 0.696 mm and 0.932 ± 0.059 for the prostate and 0.814 ± 1.002 mm and 0.801 ± 0.178 for the DILs among all 49 patients. The proposed method has shown promise for facilitating prostate and DIL delineation for DIL focal boost prostate radiation therapy.

Salvage Radiotherapy Management Decisions in Postprostatectomy Patients with Recurrent Prostate Cancer Based on 18F-Fluciclovine PET/CT Guidance

Imaging with novel PET radiotracers has significantly influenced radiotherapy decision making and radiation planning in patients with recurrent prostate cancer (PCa). The purpose of this analysis was to report the final results for management decision changes based on ¹⁸F-fluciclovine PET/CT findings and determine whether the decision change trend remained after completion of accrual. Methods: Patients with detectable prostate-specific antigen (PSA) after prostatectomy were randomized to undergo either conventional imaging (CI) only (arm A) or CI plus ¹⁸F-fluciclovine PET/CT (arm B) before radiotherapy. In arm B, positivity rates on CI and ¹⁸F-fluciclovine PET/CT for detection of recurrent PCa were determined. Final decisions on whether to offer radiotherapy and whether to include only the prostate bed or also the pelvis in the radiotherapy field were based on ¹⁸F-fluciclovine PET/CT findings. Radiotherapy decisions before and after ¹⁸F-fluciclovine PET/CT were compared. The statistical significance of decision changes was determined using the Clopper-Pearson (exact) binomial method. Prognostic factors were compared between patients with and without decision changes. Results: All 165 patients enrolled in the study had standard-of-care CI and were initially planned to receive radiotherapy. Sixty-three of 79 (79.7%) patients (median PSA, 0.33 ng/mL) who underwent ¹⁸F-fluciclovine PET/CT (arm B) had positive findings. ¹⁸F-Fluciclovine PET/CT had a significantly higher positivity rate than CI did for the whole body (79.7% vs. 13.9%; P < 0.001), prostate bed (69.6% vs. 5.1%; P < 0.001), and pelvic lymph nodes (38.0% vs. 10.1%; P < 0.001). Twenty-eight of 79 (35.4%) patients had the overall radiotherapy decision changed after ¹⁸F-fluciclovine PET/CT; in 4 of 79 (5.1%), the decision to use radiotherapy was withdrawn because of extrapelvic disease detected on ¹⁸F-fluciclovine PET/CT. In 24 of 75 (32.0%) patients with a final decision to undergo radiotherapy, the radiotherapy field was changed. Changes in overall radiotherapy decisions and radiotherapy fields were statistically significant (P < 0.001). Overall, the mean PSA at PET was significantly different between patients with and without radiotherapy decision changes (P = 0.033). Conclusion: ¹⁸F-Fluciclovine PET/CT significantly altered salvage radiotherapy decisions in patients with recurrent PCa after prostatectomy. Further analysis to determine the impact of ¹⁸F-fluciclovine PET/CT guidance on clinical outcomes after radiotherapy is in progress.

82Rubidium chloride PET discrimination of recurrent intracranial malignancy from radiation necrosis

BACKGROUND

Accurate identification and discrimination of post treatment changes from recurrent disease remains a challenge for patients with intracranial malignancies despite advances in molecular and magnetic resonance imaging. We have explored the ability of readily available Rubidium-82 chloride (82RbCl) PET to identify and distinguish progressive intracranial disease from radiation necrosis in patients previously treated with radiation therapy.

METHODS

Six patients with a total of 9 lesions of either primary (n=3) or metastatic (n=6) intracranial malignancies previously treated with stereotactic radiation surgery (SRS) and persistent contrast enhancement on MRI underwent brain 82RbCl PET imaging. Two patients with arteriovenous malformations previously treated with SRS, also had brain 82RbCl PET imaging for a total of 11 lesions studied. Histological confirmation via stereotactic biopsy/excisional resection was obtained for 9 lesions with the remaining 2 classified as either recurrent tumor or radiation necrosis based on subsequent MRI examinations. 82RbCl PET time activity curve analysis was performed which comprised lesion SUVmax, contralateral normal brain SUVmax, and tumor to background ratios (TBmax).

RESULTS

82RbCl demonstrates uptake greater than normal brain parenchyma in all lesions studied. Time activity curves demonstrated progressive uptake of 82RbCl in all lesions without evidence of washout. While recurrent disease demonstrated a greater mean SUVmax compared to radiation necrosis, no statistically significant difference between lesion SUVmax nor TBmax was found (p>0.05).

CONCLUSIONS

82RbCl PET produces high-contrast uptake of both recurrent disease and radiation necrosis compared to normal brain. However, no statistically significant difference was found between recurrent tumor and radiation necrosis.

18F-Fluciclovine Parameters on Targeted Prostate Biopsy Associated with True Positivity in Recurrent Prostate Cancer

We evaluated ¹⁸F-fluciclovine uptake parameters that correlate with true positivity for local recurrence in non-prostatectomy-treated patients. Methods: Twenty-one patients (prostate-specific antigen level, 7.4 ± 6.8 ng/mL) with biochemical recurrence after nonprostatectomy local therapy (radiotherapy and cryotherapy) underwent dual-time-point ¹⁸F-fluciclovine (364.1 ± 37.7 MBq) PET/CT from pelvis to diaphragm. Prostatic uptake over background was delineated and coregistered to a prostate-biopsy-planning ultrasound. Transrectal biopsies of ¹⁸F-fluciclovine-defined targets were completed using a 3-dimensional visualization and navigation platform. Histologic analyses of lesions were completed. Lesion characteristics including SUV_max, target-to-background ratio (TBR), uptake pattern, and subjective reader's suspicion level were compared between true-positive (malignant) and false-positive (benign) lesions. Univariate analysis was used to determine the association between PET and histologic findings. Receiver-operating-characteristic curves were plotted to determine discriminatory cutoffs for TBR. Statistical significance was set at a P value of less than 0.05. Results: Fifty lesions were identified in 21 patients on PET. Seventeen of 50 (34.0%) targeted lesions in 10 of 21 patients were positive for malignancy. True-positive lesions had a significantly higher SUV_max (6.62 ± 1.70 vs. 4.92 ± 1.27), marrow TBR (2.57 ± 0.81 vs. 1.69 ± 0.51), and blood-pool TBR (4.10 ± 1.17 vs. 2.99 ± 1.01) than false-positive lesions at the early time point (P < 0.01) and remained significant at the delayed time point, except for blood-pool TBR. Focal uptake (odds ratio, 12.07; 95% confidence interval, 2.98-48.80; P < 0.01) and subjective highest suspicion level (odds ratio, 10.91; 95% confidence interval, 1.19-99.69; P = 0.03) correlated with true positivity. Using the receiver-operating-characteristic curve, optimal cutoffs for marrow TBR were 1.9 (area under the curve, 0.82) and 1.8 (area under the curve, 0.85) at early and delayed imaging, respectively. With these cutoffs, 15 of 17 malignant lesions were identified at both time points; however, fewer false-positive lesions were detected at the delayed time point (5/33) than at the early time point (11/33). Conclusion: True positivity of ¹⁸F-fluciclovine-targeted prostate biopsy in non-prostatectomy-treated patients correlates with focal uptake, TBR (blood pool and marrow), and subjective highest suspicion level. A marrow TBR of 1.9 at the early time point and 1.8 at the delayed time point had optimal discriminating capabilities. Despite the relatively low intraprostate positive predictive value (34.0%) with ¹⁸F-fluciclovine, application of these parameters to interpretative criteria may improve true positivity in the treated prostate.

Birth preparedness and complication readiness among prenatal attendees in a teaching hospital in South West Nigeria

OBJECTIVE

To assess birth preparedness and complication readiness (BPCR) as well as knowledge of danger signs during pregnancy, labor/delivery, and the postpartum period.

METHODS

A cross-sectional study was undertaken of pregnant women attending the prenatal clinic at a tertiary hospital in Nigeria between October and December 2016. A pretested and structured questionnaire was used to collect data on BPCR, and logistic regression was performed to determine factors affecting BPCR.

RESULTS

Of 325 participants, 274 (84.3%) had knowledge of BPCR components, and 265 (81.5%) were well prepared for birth and its complications. However, only 89 (27.4%) knew key danger signs during labor/delivery and 81 (24.9%) knew those in the first 2 days after delivery. Older age, higher parity, tertiary education of women, paid employment of women and their spouses, higher social class, frequent prenatal visits, and knowledge of danger signs were significantly associated with BPCR (P<0.05). Higher parity, maternal government employment, and knowledge of danger signs during pregnancy remained determinants of BPCR on logistic regression (P<0.05).

CONCLUSION

Although there was a high level of knowledge and practice of BPCR, knowledge of key danger signs was low. Therefore, prenatal education needs to be improved with an emphasis on teaching pregnant women to recognize key danger signs.