Transfer learning supported accurate assessment of multiclass cervix type images.

Cervical cancer predominately affects women compared to lung, breast and endometrial cancer. Premature stage identification and proper treatment of this cancer may lead to 100% survival rate. The cervix type is very prominent in the detailed diagnosis of cervical cancer. High expertise and experienced gynecologist are required for an accurate diagnosis of cervical cancer. To reduce their burden, a model is proposed, based on deep learning that automatically classifies the cervix types. This paper presents Modified Deep Convolutional Neural Networks namely Modified VGG16 (MVGG16), Modified VGG19 (MVGG19), Modified ResNet50 (MRN50), Modified InceptionV3 (MIV3), and Modified InceptionResNetV2 (MIRNV2) for the classification of cervix type images. These modified networks are implemented using a Multiclass Support Vector Machine classifier. The performance metrics are tabulated and compared with pre-trained models. The simulation results show that MIRNV2 achieves the best performance compared to other models with an overall Accuracy of 92.91% and a Kappa score of 0.88. MIRNV2 model also gives better classification accuracy of 96.62% for type 1, 93.58% for type 2, and 95.61% for type 3 cervix images. Hence, this facilitates the application of MIRNV2 as a diagnostic tool to assist the gynecologist in the classification of cervix type images.

Predictors of adverse maternal outcome in jaundiced pregnant women identified as having pregnancy-specific liver disease (P-sLD).

We aimed to study the factors including model for end stage liver disease (MELD) score in predicting mortality in women with pregnancy-specific liver diseases (P-sLD). A total of 154 women with clinical jaundice were studied of which 138 women were diagnosed with P-sLD. The most common P-sLD was HELLP syndrome (51.9%) followed by acute fatty liver of pregnancy (AFLP) (17.5%). The mean age was 26.3 ± 4.7 years and the mean gestational age was 35.1 ± 4.2 weeks. The maternal death rate was 26.8% and the most common cause was coagulopathy followed by sepsis. The mean MELD score among non survivors was 25.98 ± 8.17 compared to 17.29 ± 8.12 among survivors (p value .00). On univariate analysis, gestational age at admission, presence of hypertension, the platelet count, serum creatinine, INR and MELD score were found to significant. The AUC for INR (0.82) and MELD score (0.77) was better than platelet count (0.72) and serum creatinine (0.67). On multivariate analysis, only the INR and presence of AKI were found to be significantly associated with maternal mortality. The performance of INR was better than MELD score in predicting mortality in women with P-sLD. Additional factors like platelet count may be incorporated in to MELD score for the prediction of mortality in pregnant women.IMPACT STATEMENTWhat is already known on this subject? Pregnancy-specific liver disorders (P-sLD) have significant effect on maternal and foetal outcome, often considered as a spectrum of disease with significant overlap of clinical and laboratory parameters. MELD score is used reliably outside the pregnancy to predict mortality may not be good in pregnant women. There are only few studies that looked at the factors predictive of adverse maternal outcome.What do the results of this study add? Though we have demonstrated that MELD score was significantly high among non-survivors, serum bilirubin an important component of MELD score was not found to be significant. The other factors which were found to be significant on univariate analysis include gestational age at admission, hypertension and platelet count. However, international normalised ratio (INR) and acute kidney injury (AKI) were the factors independently associated with mortality.What are the implications for clinical practice and/or further research? The utility of MELD score in P-sLD should be studied prospectively in different populations. Moreover, the feasibility of developing a simple model which incorporates platelet count in addition to other components of MELD score should also be explored in future studies.