Improved Prediction of Ovarian Cancer Using Ensemble Classifier and Shaply Explainable AI

The importance of detecting and preventing ovarian cancer is of utmost significance for women's overall health and wellness. Referred to as the "silent killer," ovarian cancer exhibits inconspicuous symptoms during its initial phases, posing a challenge for timely identification. Identification of ovarian cancer during its advanced stages significantly diminishes the likelihood of effective treatment and survival. Regular screenings, such as pelvic exams, ultrasound, and blood tests for specific biomarkers, are essential tools for detecting the disease in its early, more treatable stages. This research makes use of the Soochow University ovarian cancer dataset, containing 50 features for the accurate detection of ovarian cancer. The proposed predictive model makes use of a stacked ensemble model, merging the strengths of bagging and boosting classifiers, and aims to enhance predictive accuracy and reliability. This combination harnesses the benefits of variance reduction and improved generalization, contributing to superior ovarian cancer prediction outcomes. The proposed model gives 96.87% accuracy, which is currently the highest model result obtained on this dataset so far using all features. Moreover, the outcomes are elucidated utilizing the explainable artificial intelligence method referred to as SHAPly. The excellence of the suggested model is demonstrated through a comparison of its performance with that of other cutting-edge models.

Ultrabright NIR-II Emissive Polymer Dots for Metastatic Ovarian Cancer Detection

Intraoperative diagnosis of metastatic tumors is of significant importance to the treatment of ovarian cancer. NIR-II fluorescence imaging holds great promise for facile detection of tumor in situ with high sensitivity and resolution. Herein, a kind of NIR-II fluorescent polymer dots (NIR-II Pdots) with high brightness is developed for real-time detection of metastatic ovarian cancer via NIR-II fluorescence imaging. The NIR-II Pdots are constructed via the self-assembly of NIR-II emissive aggregation induced emission luminogens (NIR-II AIEgens) and poly (styrene)-graft-poly(ethylene glycol) in water. Such NIR-II Pdots show very high fluorophore contents of nearly 30% and high quantum yield of 5.4% at emission maximum near 1020 nm. Further modification of the NIR-II Pdots with targeting peptides yields NIR-II Pdots-GnRH, which can afford enhanced affinity of NIR-II Pdots to ovarian cancer. Upon intravenous injection of the NIR-II Pdots, whole-body organs and vessels, peritoneal and lymphatic metastases of ovarian cancer are clearly visualized by NIR-II fluorescence imaging. Under the guidance of NIR-II fluorescence imaging, the metastatic foci with the diameter down to ≈2 mm can be facilely eliminated. The results indicate preclinical potential value of the NIR-II Pdots for metastatic ovarian cancer detection.

Point-of-care Ratiometric Fluorescence Imaging of Tissue for the Diagnosis of Ovarian Cancer

During a minimally invasive tumor resection procedure, it is still a challenge to rapidly and accurately trace tiny malignant tumors in real time. Fluorescent molecular imaging is considered an efficient method of localizing tumors during surgery due to its high sensitivity and biosafety. On the basis of the fact that γ-glutamyltranspeptidase (GGT) is overexpressed in ovarian cancer, we herein designed a highly sensitive ratiometric fluorescent GGT-responsive probe Py-GSH for rapid tumor detection.

Methods: The GGT response probe (Py-GSH) was constructed by using GSH group as a response group and pyrionin B as a fluorescent reporter. Py-GSH was characterized for photophysical properties, response speed and selectivity of GGT and response mechanism. The anti-interference ability of ratiometric probe Py-GSH to probe concentration and excitation power was evaluated both in vitro and in tissue. The biocompatibility and toxicity of the ratiometric probe was examined using cytoxicity test. The GGT levels in different lines of cells were determined by ratiometric fluorescence imaging and cytometry analysis.

Results: The obtained probe capable to rapidly monitored GGT activity in aqueous solution with 170-fold ratio change. By ratiometric fluorescence imaging, the probe Py-GSH was also successfully used to detect high GGT activity in solid tumor tissues and small peritoneal metastatic tumors (~1 mm in diameter) in a mouse model. In particular, this probe was further used to determine whether the tissue margin following clinical ovarian cancer surgery contained tumor.

Conclusion: In combination of ratiometric fluorescence probes with imaging instrument, a point-of-care imaging method was developed and may be used for surgical navigation and rapid diagnosis of tumor tissue during clinical tumor resection.

Tuning to optimize SVM approach for assisting ovarian cancer diagnosis with photoacoustic imaging

Support vector machine (SVM) is one of the most effective classification methods for cancer detection. The efficiency and quality of a SVM classifier depends strongly on several important features and a set of proper parameters. Here, a series of classification analyses, with one set of photoacoustic data from ovarian tissues ex vivo and a widely used breast cancer dataset- the Wisconsin Diagnostic Breast Cancer (WDBC), revealed the different accuracy of a SVM classification in terms of the number of features used and the parameters selected. A pattern recognition system is proposed by means of SVM-Recursive Feature Elimination (RFE) with the Radial Basis Function (RBF) kernel. To improve the effectiveness and robustness of the system, an optimized tuning ensemble algorithm called as SVM-RFE(C) with correlation filter was implemented to quantify feature and parameter information based on cross validation. The proposed algorithm is first demonstrated outperforming SVM-RFE on WDBC. Then the best accuracy of 94.643% and sensitivity of 94.595% were achieved when using SVM-RFE(C) to test 57 new PAT data from 19 patients. The experiment results show that the classifier constructed with SVM-RFE(C) algorithm is able to learn additional information from new data and has significant potential in ovarian cancer diagnosis.

Recognition algorithm for assisting ovarian cancer diagnosis from coregistered ultrasound and photoacoustic images: ex vivo study

Unique features and the underlining hypotheses of how these features may relate to the tumor physiology in coregistered ultrasound and photoacoustic images of ex vivo ovarian tissue are introduced. The images were first compressed with wavelet transform. The mean Radon transform of photoacoustic images was then computed and fitted with a Gaussian function to find the centroid of a suspicious area for shift-invariant recognition process. Twenty-four features were extracted from a training set by several methods, including Fourier transform, image statistics, and different composite filters. The features were chosen from more than 400 training images obtained from 33 ex vivo ovaries of 24 patients, and used to train three classifiers, including generalized linear model, neural network, and support vector machine (SVM). The SVM achieved the best training performance and was able to exclusively separate cancerous from non-cancerous cases with 100% sensitivity and specificity. At the end, the classifiers were used to test 95 new images obtained from 37 ovaries of 20 additional patients. The SVM classifier achieved 76.92% sensitivity and 95.12% specificity. Furthermore, if we assume that recognizing one image as a cancer is sufficient to consider an ovary as malignant, the SVM classifier achieves 100% sensitivity and 87.88% specificity.