Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks

Abstract

The purpose of this study was to develop a method of classifying cancers to specific diagnostic categories based on their gene expression signatures using artificial neural networks (ANNs). We trained the ANNs using the small, round blue-cell tumors (SRBCTs) as a model. These cancers belong to four distinct diagnostic categories and often present diagnostic dilemmas in clinical practice. The ANNs correctly classified all samples and identified the genes most relevant to the classification. Expression of several of these genes has been reported in SRBCTs, but most have not been associated with these cancers. To test the ability of the trained ANN models to recognize SRBCTs, we analyzed additional blinded samples that were not previously used for the training procedure, and correctly classified them in all cases. This study demonstrates the potential applications of these methods for tumor diagnosis and the identification of candidate targets for therapy.

Identification of expressed genes characterizing long-term survival in malignant glioma patients

“…Several works (Sallinen et al, 2000;Khan et al, 2001;Ramaswamy et al, 2001;Rickman et al, 2001;Agrawal et al, 2002;Kim et al, 2002;Veer et al, 2002;Vijver et al, 2002;Boom et al, 2003;Godard et al, 2003;Hunter et al, 2003;Mischel et al, 2003;Nutt et al, 2003;Shai et al, 2003;Sorlie et al, 2003;Freije et al, 2004;Mischel et al, 2004;Hoelzinger et al, 2005;Liang et al, 2005;Nigro et al, 2005;Rich et al, 2005;Somasundaram et al, 2005;Wong et al, 2005) showed the usefulness of utilizing methods of analysis of multiple forms of data including both clinical and multiple genes, to achieve a more precise discrimination of outcomes for individual patients. The same logical use of multiple forms of data and methods of analysis has been applied in the present study to accurately achieve better classification and prediction of glioma patients.…”

“…Recently, developed microarray technology has permitted development of multi-organ cancer classification including gliomas (Ramaswamy et al, 2001;Rickman et al, 2001;Kim et al, 2002;Hunter et al, 2003;Mischel et al, 2004), identification of tumor subclasses (Khan et al, 2001;Mischel et al, 2003;Shai et al, 2003;Sorlie et al, 2003;Liang et al, 2005;Nigro et al, 2005;Wong et al, 2005), discovery of progression markers (Sallinen et al, 2000;Agrawal et al, 2002;van de Boom et al, 2003;Godard et al, 2003;Hoelzinger et al, 2005;Rich et al, 2005;Somasundaram et al, 2005) and prediction of disease outcomes (van't Veer et al, 2002;van de Vijver et al, 2002;Nutt et al, 2003;Freije et al, 2004). Unlike clinicopathological staging, molecular staging can predict long-term outcomes of any individual based on gene expression profile of the tumor at diagnosis.…”

The Application of Artificial Intelligence to Microarray Data: Identification of a Novel Gene Signature to Identify Bladder Cancer Progression

“…Artificial intelligence (AI) is a machine learning approach without these prerequisites. Various AI techniques exist [8] and successful microarray analysis has been reported using artificial neural networks (ANN) [9] [10] and support vector machines (SVMs) [11,12] in non-urothelial malignancies. However, the hidden working layer of an ANN prevents model understanding and hinders its acceptance by the scientific community [13], whilst SVMs still use proximity to infer class-gene associations and function poorly with respect to interpretability [14].…”

Array CGH and gene-expression profiling reveals distinct genomic instability patterns associated with DNA repair and cell-cycle checkpoint pathways in Ewing's sarcoma

“…Previous works have focused their research on array expression profiles of ES cell lines and primary tumours in order to identify tumour classification profiles, new EWS/FLI1 targets or to reveal distinct expression signatures associated to different outcomes (Khan et al, 2001;Ohali et al, 2004;Bandres et al, 2005;Mendiola et al, 2006). A more recent genomic approach is the use of array-based comparative genomic hybridization (aCGH), which has been successfully used for the detection of genomic imbalances in human tumours (Pinkel and Albertson, 2005) in a most precise manner than chromosome-based conventional techniques.…”