Lung Cancer is a leading cause of human loss globally i.e. compared with other cancer relatedrndeaths. The five-year relative survival rate of lung cancer is only 16%; however, earlyrnrecognition of nodules and proper treatment of this disease reduce the death rate due to lungrncancer up to 20%. To detect such nodules CT lung image analysis has been used by radiologistsrnall over the world. However, analysis of these images is a very challenging task for radiologists,rnbecause the number of slices in one scan can be up to 600. Therefore, computer aided-detectionrn(CAD) systems are very important for a quicker and more precise assessment of the data. Falsernpositive reduction is one of a vital element of computer aided diagnosis (CAD system), whichrnplays an important role in lung cancer diagnosis and early treatment. In this thesis we proposedrnto design a framework for classification of candidate CT image slices by employing 3DrnConvolutional Neural Network (CNNs) to reduce a significant number of false positiverncandidates. 3D CNNs are favorable than 2D CNNs because 3D CNNs can encode richer spatialrninformation and extract more representative features via their structural architecture trained withrn3D samples. The proposed design has been extensively validated by using the dataset obtainedrnfrom LUNA16 challenge providers. rnThe proposed approach mainly consists of three steps Pre-processing, Feature extraction rn& Classification and Fusion. In the preprocessing phase we carefully examined our data set andrnperform resampling to avoid image slice thickness variations because of different CT machines.rnIn addition to that we employed data augmentation to reduce class imbalance between groundrntruth nodules and false positive candidates. Sizes of the nodules varied from 3mm up to 30mm,rnso we extract four receptive fields to encompass all nodule types (i.e. small, medium and largernnodules), aiming to understand the effect of input patch sizes in the performance of the system.rnWe designed four 3D CNNs for the corresponding four patch sizes, each CNNs model containsrnfrom 3 convolutional layers. We employed model fusion technique to acquire an improved resultrnby using the aggregate strengths of each model. The proposed framework has been tested byrnthe dataset provided by the LUNA16 Challenge and we achieved the competitionrnperformance metric (CPM) score of 0.8541 with a highest sensitivity of 0.8706 with 1 falsernpositive per scan and 0.9275 at 8 false positives per scan.Generally, from the test results we noticed that increasing the input patch size increases thernoverall CPM score because larger patch sizes can be able to encompass a large number ofrnnodules within the dataset and any reasonable fusion method can be able to boost the overallrnclassification performance.