Algorithm-assisted diagnosis of Hirschsprung's disease - evaluation of robustness and comparative image analysis on data from various labs and slide scanners.

BACKGROUND: Differences in the preparation, staining and scanning of digital pathology slides create significant pre-analytic variability. Algorithm-assisted tools must be able to contend with this variability in order to be applicable in clinical practice. In a previous study, a decision support algorithm was developed to assist in the diagnosis of Hirschsprung's disease. In the current study, we tested the robustness of this algorithm while assessing for pre-analytic factors which may affect its performance. METHODS: The decision support algorithm was used on digital pathology slides obtained from four different medical centers (A-D) and scanned by three different scanner models (by Philips, Hamamatsu and 3DHISTECH). A total of 192 cases and 1782 slides were used in this study. RGB histograms were constructed to compare images from the various medical centers and scanner models and highlight the differences in color and contrast. RESULTS: The algorithm was able to correctly identify ganglion cells in 99.2% of cases, from all medical centers (All scanned by the Philips slide scanner) as well as 95.5% and 100% of the slides scanned by the 3DHISTECH and Hamamatsu brand slide scanners, respectively. The total error rate for center D was lower than the other medical centers (3.9% vs 7.1%, 10.8% and 6% for centers A-C, respectively), the vast majority of errors being false positives (3.45% vs 0.45% false negatives). The other medical centers showed a higher rate of false negatives in relation to false positives (6.81% vs 0.29%, 9.8% vs 1.2% and 5.37% vs 0.63% for centers A-C, respectively). The total error rates for the Philips, Hamamatsu and 3DHISTECH brand scanners were 3.9%, 3.2% and 9.8%, respectively. RGB histograms demonstrated significant differences in pixel value distribution between the four medical centers, as well as between the 3DHISTECH brand scanner when compared to the Philips and Hamamatsu brand scanners. CONCLUSIONS: The results reported in this paper suggest that the algorithm-based decision support system has sufficient robustness to be applicable for clinical practice. In addition, the novel method used in its development - Hierarchial-Contexual Analysis (HCA) may be applicable to the development of algorithm-assisted tools in other diseases, for which available datasets are limited. Validation of any given algorithm-assisted support system should nonetheless include data from as many medical centers and scanner models as possible.

Designing robots that do no harm: understanding the challenges of Ethics for Robots.

This article describes key challenges in creating an ethics "for" robots. Robot ethics is not only a matter of the effects caused by robotic systems or the uses to which they may be put, but also the ethical rules and principles that these systems ought to follow-what we call "Ethics for Robots." We suggest that the Principle of Nonmaleficence, or "do no harm," is one of the basic elements of an ethics for robots-especially robots that will be used in a healthcare setting. We argue, however, that the implementation of even this basic principle will raise significant challenges for robot designers. In addition to technical challenges, such as ensuring that robots are able to detect salient harms and dangers in the environment, designers will need to determine an appropriate sphere of responsibility for robots and to specify which of various types of harms must be avoided or prevented. These challenges are amplified by the fact that the robots we are currently able to design possess a form of semi-autonomy that differs from other more familiar semi-autonomous agents such as animals or young children. In short, robot designers must identify and overcome the key challenges of an ethics for robots before they may ethically utilize robots in practice.

Short Training Significantly Improves Ganglion Cell Detection Using an Algorithm-Assisted Approach.

CONTEXT.­: Medical education in pathology relies on the accumulation of experience gained through inspection of numerous samples from each entity. Acquiring sufficient teaching material for rare diseases, such as Hirschsprung disease (HSCR), may be difficult, especially in smaller institutes. The current study makes use of a previously developed decision support system using a decision support algorithm meant to aid pathologists in the diagnosis of HSCR. OBJECTIVE.­: To assess the effect of a short training session on algorithm-assisted HSCR diagnosis. DESIGN.­: Five pathologists reviewed a data set of 568 image sets (1704 images in total) selected from 50 cases by the decision support algorithm and were tasked with scoring the images for the presence or absence of ganglion cells. The task was repeated a total of 3 times. Each pathologist had to complete a short educational presentation between the second and third iterations. RESULTS.­: The training resulted in a significantly increased rate of correct diagnoses (true positive/negative) and a decreased need for referrals for expert consultation. No statistically significant changes in the rate of false positives/negatives were detected. CONCLUSIONS.­: A very short (<10 minutes) training session can greatly improve the pathologist's performance in the algorithm-assisted diagnosis of HSCR. The same approach may be feasible in training for the diagnosis of other rare diseases.

Automatic ganglion cell detection for improving the efficiency and accuracy of hirschprung disease diagnosis.

Histopathologic diagnosis of Hirschsprung's disease (HSCR) is time consuming and requires expertise. The use of artificial intelligence (AI) in digital pathology is actively researched and may improve the diagnosis of HSCR. The purpose of this research was to develop an algorithm capable of identifying ganglion cells in digital pathology slides and implement it as an assisting tool for the pathologist in the diagnosis of HSCR. Ninety five digital pathology slides were used for the construction and training of the algorithm. Fifty cases suspected for HSCR (727 slides) were used as a validation cohort. Image sets suspected to contain ganglion cells were chosen by the algorithm and then reviewed and scored by five pathologists, one HSCR expert and 4 non-experts. The algorithm was able to identify ganglion cells with 96% sensitivity and 99% specificity (in normal colon) as well as to correctly identify a case previously misdiagnosed as non-HSCR. The expert was able to achieve perfectly accurate diagnoses based solely on the images suggested by the algorithm, with over 95% time saved. Non-experts would require expert consultation in 20-58% of the cases to achieve similar results. The use of AI in the diagnosis of HSCR can greatly reduce the time and effort required for diagnosis and improve accuracy.

Geographic Heterogeneity for Mismatch Repair Proteins Is Associated with Defects in DNA Repair.

BACKGROUND: Evaluation of mismatch repair (MMR) deficiency is conducted via immunohistochemistry or by microsatellite instability (MSI) analysis. Heterogeneous immunohistochemistry staining for MMR proteins may show different patterns; however, according to current guidelines, all of those patterns should be interpreted as MMR proficient. This conclusion might lead to false negative results because although most cases of heterogeneity stem from technical factors and biological variability, other types of heterogeneity represent true MMR deficiency. OBJECTIVES: To identify a unique heterogeneity pattern that is associated with true MMR loss. METHODS: We analyzed 145 cases of colorectal carcinoma. Immunohistochemistry staining for MLH1, PMS2, MSH2, and MSH6 were performed. We defined geographic heterogeneity as areas of tumor nuclear staining adjacent to areas of loss of tumor nuclear staining with intact staining in the surrounding stroma. All cases were evaluated for the presence of geographic heterogeneity. In addition, 24 cases were also evaluated by MSI testing. RESULTS: Of the 145 cases, 24 (16.5%) were MMR deficient. Of the 24 cases for which MSI analysis was also available, 10 cases (41.7%) demonstrated biological heterogeneity, 5 (20.8%) demonstrated technical heterogeneity, and 2 (8.3%) demonstrated geographic heterogeneity. Only the two cases with geographic heterogeneity were MSI-high via MSI analysis. In addition, a germline mutation in MSH-6 was identified in one of these cases. CONCLUSIONS: Geographic heterogeneity may raise a suspicion for a MMR-deficient case, which should be further analyzed using additional methodologies such as MSI analysis.

Can We Prevent Stillbirth?

OBJECTIVE: This study aimed to identify the frequency of potentially preventable causes of stillbirth in a large heterogeneous population. STUDY DESIGN: This is a retrospective study of all stillbirth cases between January 2011 and December 2016 at a single tertiary medical center. Deliveries resulting from a nonviable fetus prior to 24 weeks of gestation, intrapartum fetal death, and incomplete stillbirth workup were excluded. Potentially preventable stillbirth was defined as that of a nonanomalous fetus that most likely resulted from one or more of the following: (1) placental-mediated complications, (2) postterm pregnancy, (3) monochorionicity-associated complications, (4) cholestasis of pregnancy, (5) preventable or treatable infections, and (6) isoimmunization. RESULTS: During the study period, 312 stillbirths were identified, 228 of which met the inclusion criteria. Of the 110 cases with a recognized cause, 47 (20.6%) were potentially preventable. The most common causes were placental-mediated complications and preventable or treatable infections, accounting for 75 and 9% of all potentially preventable causes, respectively. There were no recognizable maternal risk factors for potentially preventable stillbirth. CONCLUSION: One-fifth of all causes of stillbirth are potentially preventable. Due to the significant contribution of placental-mediated complications to preventable stillbirth, close sonographic surveillance and timely delivery may decrease risk substantially.

Surgery for Recurrent High-Grade Glioma After Treatment with Bevacizumab.

BACKGROUND: Bevacizumab (BVZ) is an antiangiogenic agent approved by the Food and Drug Administration that is used for the treatment of recurrent glioblastoma. Complications related to impaired healing may adversely affect patients resected for recurrent high-grade glioma (HGG) after treatment with BVZ. OBJECTIVE: To examine the complication rate, outcome, and tumor vasculature in patients resected for recurrent HGG after treatment with BVZ. METHODS: Data were reviewed retrospectively from patients undergoing surgery for recurrent HGG after treatment with BVZ. Results were compared with a control group of recurrently operated BVZ-naïve HGG. Tumor samples and magnetic resonance imaging scans were analyzed. RESULTS: Fifteen patients underwent HGG resection after progression after BVZ. Forty-four BVZ-naïve patients who underwent surgeries for tumor recurrence were included as controls. Median time from BVZ treatment to surgery was 30 days (2-107). Median overall survival from time of tumor diagnosis was 21.0 months (12-83.0), and median survival from post-BVZ surgery was 5.0 months (2.0-19.0), compared with 8.1 months in BVZ-naïve controls measured from time of their last reoperation. Five of the 15 patients survived 6 or more months after post-BVZ surgery. Nine patients developed postsurgical complications requiring intervention. Complication rates for surgery after BVZ treatment were 66.7% compared with 38.6% in the control group (P = 0.077). We did not see overt changes in histopathology or immunohistochemistry staining; however, tumor vasculature in tumors resected after treatment with BVZ showed a significant decrease in mean vessel density. CONCLUSIONS: Surgery for recurrent HGG may be feasible in a select group of patients. Mean tumor vessel density may be decreased after treatment with BVZ.

Correlation between the presence of KRAS mutation and the morphometric characteristics of colorectal carcinoma cell nuclei.

OBJECTIVE: To determine whether there is an association between KRAS status and changes in nuclear morphometric properties. STUDY DESIGN: A total of 60 cases were included in the analysis. Thirty cases carried a mutated KRAS oncogene, while 30 bore a wild-type KRAS gene. Computerized morphometry was used to measure various nuclear indices of size, shape, and texture. RESULTS: Nuclear ellipticity (p = 0.02) and chromatin textural contrast (p = 0.02) were both significantly higher in the mutated KRAS group. However, chromatin contrast was the only independent predictor of KRAS status. A ROC analysis detected a best chromatin contrast cutoff point of 798, with a sensitivity of 89.7% and a specificity of 52%. Tumor grade was also associated with nuclear ellipticity (p = 0.001). Yet, no association was found between tumor grade and KRAS status. CONCLUSION: Our results suggest that the presence of a KRAS mutation in colorectal carcinoma is associated with alterations in the nuclear morphometric properties of contrast and ellipticity.