Modified fuzzy clustering with partial supervision algorithm in classification and recognition of pulmonary nodules

Pulmonary nodule risk classification in adenocarcinoma from CT images using deep CNN with scale transfer module

“…Radiomics and DL methods are usually established to calculate different kinds of features to predict clinical cancer status. There are lots of work to screen pulmonary nodule into binary categories as benign and malignancy [9–11, 22–24]. For instance, Jacobs et al .…”

“…In the past several years, many studies on the pulmonary nodule classification are based on the benign and malignant two‐classification algorithm [5–11]. In fact, lung cancer generally has a long growth process of substitution, migration, evolution and transformation, so pulmonary nodule can be classified more detailed.…”

“…However, the non‐uniform distribution of pulmonary nodule size in datasets also affects the classification accuracy. Li et al [10] applied an improved semi‐supervised FCM clustering algorithm to cluster 11 features including grey variance, boundary roughness and moment invariant, and accuracy reached 77.6% on the Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI) dataset. Alahmari et al [11] combined delta radiomics with conventional (non‐delta) features, using machine learning to classify the benign and malignant pulmonary nodules, found the best improved average area under the receiver operating characteristic curve (AUC) was 0.822 when delta features were similar with conventional features versus an AUC 0.773 for conventional features only.…”