For Project Enquiry +91 9791 044 044

EXISTING SYSTEM

PROPOSED SYSTEM

EXISTING CONCEPT:

          Computed tomography (CT) is the most accurate imaging modality to obtain anatomical information about lung nodules.

          The filter based feature extraction techniques, such as maximum response (MR8), are also widely applied in second level and Latent Semantic Analysis is used.

PROPOSED CONCEPT:

          A novel image classification method for the four common types of lung nodule is combining of patch-based image representation, and then feature set patch description, and contextual latent semantic analysis-based classifier to calculate the probabilistic estimations.

EXISTING TECHNIQUE:

          Local Density Maximum (LDM)

PROPOSED TECHNIQUE:

         SVM (Support vector machine          PLSA(problematic latent semantic analysis)

TECHNIQUE DEFINITION:

          The LDM algorithm begins to threshold it with an initial threshold value that can be the maximal density value of the profile.

TECHNIQUE DEFINITION:

          SVM are supervised learning models with associated learning algorithms that analyze data and recognize patterns, used or classification and regression analysis.

          LSA for certain level are extracted by pLSA. Assuming there are M images and the dictionary size is N, for each level-l, we could obtain the dataset of M levels.

DRAWBACKS:

          Direct classification from these would still be problematic.

          Contextual information surrounding the lung nodules could be incorporated to improve nodule classification is complicated segmentation process.

ADVANTAGES:

          Super pixel formulation dividing an image into multiple segments, and reduce spurious labeling due to noise

          To overcome the problem of the lung nodule overlapping adjacent structures.