Presentation and study of robustness for several methods to classify individuals based on their gene expressions

Several studies have shown that it is possible to detect cancer tissues based on gene expressions using methods of machine learning. The main problem with classifying gene expression data is to obtain accurate rules that are easy to interpret and provide indications for follow up studies. Indeed high accuracy is hard to achieve due to the small number of observations and the large amount of genes in the human genome. Some methods of machine learning are based on an important quantity of genes, which lead to decision rules that are usually difficult to interpret. These methods were tested on different samples and their results were compared. Most of them provided good results with a high accuracy. Among these methods for gene classification one distanced itself from the others by producing transparents results which were readily interpretable and were very useful for follow up studies. It highlighted pair of genes that were the most efficient to classify individuals with respect to their gene expressions. This is the so called Top Scoring Pair (TSP) classifier. This method achieves prediction rates that are as high as those of the other methods. In contrast to other classifiers which use considerably more genes and more complicated procedures, the TSP has an easy and quick implementation and involves very few genes, namely only two. This provides very easy rules that are accurate and transparent. Finally, the TSP is paramter-free, which avoids overfitting and inflation of the estimation of the prediction rate.

Prediction of Survival and Risk Assessment Using Joint Analysis of Microarray Gene Expression Data

Haleh Yasrebi

Gene expression profiles have been widely used in molecular classification, diagnosis and prediction, particularly in the area of oncology where accurate and early diagnosis is needed for appropriate treatment. Avoiding under-/over-treatment when it is not necessary can extend a patient's survival and prevent disease recurrence. These high-throughput assay technologies have generated terabytes of data exploited extensively to provide insights on cancer biology and the underlying mechanism of disease progression. The ultimate goal is to identify possibly tailored treatment and therapy for personalized medicine. Analysis of microarray data is constrained by the following characteristics: (i) noisy due to missing or erroneous values; (ii) high dimensional due to a large number of genes versus a few number of samples in which their expression levels are measured; (iii) costly due to expensive microarray experiments. Abundant microarray gene expression data should be processed by appropriate computational and statistical learning methodologies such as machine learning techniques. These methods are robust to noisy data and have a great capacity to analyze high dimensional data. Their computational power is nevertheless limited to sample size based on which these methods are built. These algorithms have been widely applied to microarray gene expression data to identify a set of genes known as a gene signature whose expressions are highly correlated to a target value or outcome such as disease status, tumor subtype, a patient's survival time, risk of mortality or cancer relapse. Prediction of survival time and a patient's risk which is unknown at diagnosis presents a more challenging task for machine learning methods than tumor subtype or disease classification, which is already established by oncologists. The properties of microarray data cited above, the limitation of the number of samples in cancer patients and dependency of the machine learning methods' performance on sample size justify joint analysis of microarray data to increase the number of samples. We applied joint analysis methods to breast and lung cancer data sets to improve survival prediction and risk assessment. In overall, no significant improvement or deterioration of the performance accuracy was obtained with joint analysis. However, increasing sample size helped to identify robust or stable gene signatures predictive of survival time and risk assessment. Our achievements and learned-lessons from joint analysis of microarray gene expression data can be used as a guideline for future research studies in classification and prediction.

EPFL2010

Can survival prediction be improved by merging gene expression data sets?

Philipp Bucher, Haleh Yasrebi

BACKGROUND: High-throughput gene expression profiling technologies generating a wealth of data, are increasingly used for characterization of tumor biopsies for clinical trials. By applying machine learning algorithms to such clinically documented data sets, one hopes to improve tumor diagnosis, prognosis, as well as prediction of treatment response. However, the limited number of patients enrolled in a single trial study limits the power of machine learning approaches due to over-fitting. One could partially overcome this limitation by merging data from different studies. Nevertheless, such data sets differ from each other with regard to technical biases, patient selection criteria and follow-up treatment. It is therefore not clear at all whether the advantage of increased sample size outweighs the disadvantage of higher heterogeneity of merged data sets. Here, we present a systematic study to answer this question specifically for breast cancer data sets. We use survival prediction based on Cox regression as an assay to measure the added value of merged data sets. RESULTS: Using time-dependent Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) and hazard ratio as performance measures, we see in overall no significant improvement or deterioration of survival prediction with merged data sets as compared to individual data sets. This apparently was due to the fact that a few genes with strong prognostic power were not available on all microarray platforms and thus were not retained in the merged data sets. Surprisingly, we found that the overall best performance was achieved with a single-gene predictor consisting of CYB5D1. CONCLUSIONS: Merging did not deteriorate performance on average despite (a) The diversity of microarray platforms used. (b) The heterogeneity of patients cohorts. (c) The heterogeneity of breast cancer disease. (d) Substantial variation of time to death or relapse. (e) The reduced number of genes in the merged data sets. Predictors derived from the merged data sets were more robust, consistent and reproducible across microarray platforms. Moreover, merging data sets from different studies helps to better understand the biases of individual studies and can lead to the identification of strong survival factors like CYB5D1 expression.

Public Library of Science2009

Alternative search techniques for face detection using location estimation and binary features

Venkatesh Bala Subburaman

The sliding window approach is the most widely used technique to detect objects from an image. In the past few years, classifiers have been improved in many ways to increase the scanning speed. Apart from the classifier design (such as the cascade), the scanning speed also depends on a number of different factors (such as grid spacing, and scale at which the image is searched). Scanning grid spacing controls the number of subwindows being processed, thus controlling the speed of detection.When the scanning grid spacing is larger than the tolerance of the trained classifier it can suffer from low detections. In this thesis, we propose an alternative search technique, which can improve the detections when lesser number of subwindows are processed. First, we present a technique to reduce the number of miss detections while increasing the grid spacing when using the sliding window approach for object detection. This is achieved by using a small patch to predict the location of an object within a local search area. To achieve speed, it is necessary that the time taken for location prediction is comparable or better than the time it takes in average for the object classifier to reject a subwindow. We use binary features and a decision tree as it proved to be efficient for our application. In the process we also propose a variation of an existing binary feature (Ferns) with similar performance, and requires only half the number of pixel access when compared to Fern feature. We analyze the effect of patch size on location estimation and also evaluate our approach on several face databases. Experimental evaluation shows better detection rate and speed with our proposed approach for larger grid spacing (lesser number of subwindows) when compared to standard scanning technique. We also show that by using a simple interest point detector based on quantized gradient orientation, as the front-end to the proposed location estimation technique, we can achieve better performance even when fewer number of subwindows are processed. The interest points detected can be assumed as a non-regular grid compared to regular grid in the sliding window framework. A few image patches are sampled around an interest point for estimating the probable face location and further verified using a strong face classifier. Experiment results show that using an interest point detector can reduce the number of subwindows processed while maintaining a good detection rate.

ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE2012