This paper presents a probability segmentation algorithm for lung nodules based on three-dimensional features. Firstly, we computed intensity and texture features in region of interest (ROI) pixel by pixel to get their feature vector, and then classified all the pixels based on their feature vector. At last, we carried region growing on the classified result, and got the final segmentation result. Using the public Lung Imaging Database Consortium (LIDC) lung nodule datasets, we verified the performance of proposed method by comparing the probability map within LIDC datasets, which was drawn by four radiology doctors separately. The experimental results showed that the segmentation algorithm using three-dimensional intensity and texture features would be effective.
With the wide utilization of high-resolution computed tomography (HRCT) in the lung cancer screening, patients detected with pulmonary ground-glass nodules (GGNs) have increased over time and account for a large proportion of all thoracic diseases. Because of its less invasiveness and fast recovery, video-assisted thoracoscopic surgery (VATS) is currently the first choice of surgical approach to lung nodule resection. However, GGNs are usually difficult to recognize during VATS, and failure of nodule localization would result in conversion to thoracotomy or extended lung resection. In order to cope with this problem, a series of approaches for pulmonary nodule localization have developed in the last few years. This article aims to summarize the reported methods of lung nodule localization and analyze its corresponding pros and cons, in order to help thoracic surgeons to choose appropriate localization method in different clinical conditions.
Peripheral pulmonary lesions (PPLs) are generally considered as lesions in the peripheral one-third of the lung. A computed tompgraphy (CT) guided transthoracic needle aspiration/biopsy or transbronchial approach using a bronchoscope has been the most generally accepted methods. Navigation technique can effectively improve the diagnosis rate of peripheral pulmonary lesions, reduce the incidence of complications, shorten the time of diagnosis, and make the patients get timely and effective treatment.
A method was proposed to detect pulmonary nodules in low-dose computed tomography (CT) images by two-dimensional convolutional neural network under the condition of fine image preprocessing. Firstly, CT image preprocessing was carried out by image clipping, normalization and other algorithms. Then the positive samples were expanded to balance the number of positive and negative samples in convolutional neural network. Finally, the model with the best performance was obtained by training two-dimensional convolutional neural network and constantly optimizing network parameters. The model was evaluated in Lung Nodule Analysis 2016(LUNA16) dataset by means of five-fold cross validation, and each group's average model experiment results were obtained with the final accuracy of 92.3%, sensitivity of 92.1% and specificity of 92.6%.Compared with other existing automatic detection and classification methods for pulmonary nodules, all indexes were improved. Subsequently, the model perturbation experiment was carried out on this basis. The experimental results showed that the model is stable and has certain anti-interference ability, which could effectively identify pulmonary nodules and provide auxiliary diagnostic advice for early screening of lung cancer.
ObjectiveTo introduce the application of mixed reality technique to the preoperative and intraoperative pulmonary nodules surgery.MethodsOne 49-year female patient with multiple nodules in both lobes of the lung who finally underwent uniportal thoracoscopic resection of superior segment of left lower lobe and wedge resection of left upper lobe was taken as an example. The Mimics medical image post-processing software was used to reconstruct the patient's lung image based on the DICOM data of the patient's chest CT image before the surgery. The three-dimensional reconstructed image data was imported into the HoloLens glasses, and the preoperative discussions were conducted with the assistance of mixed reality technology to formulate the surgical methods, and the preoperative conversation with the patients was also conducted. At the same time, mixed reality technology was used to guide the surgery in real time.ResultsMixed reality technology can clearly pre-show the important anatomical structures of blood vessels, trachea, lesions and their positional relationship. With the help of mixed reality technology, the operation went smoothly. The total operation time was 49 min, the precise dorsal resection time was 27 min, and the intraoperative blood loss was about 39 mL. The patient recovered well and was discharged from hospital smoothly after surgery.ConclusionMixed reality technology has certain application value before and during the surgery for pulmonary nodules. The continuous maturity of this technology and its further application in clinics will not only bring a new direction to the development of thoracic surgery, but also provide a wide prospect.
ObjectiveTo evaluate the effectiveness of the artificial intelligence-assisted diagnosis and treatment system in distinguishing benign and malignant lung nodules and the infiltration degree.MethodsClinical data of 87 patients with pulmonary nodules admitted to the First Affiliated Hospital of Xiamen University from January 2019 to August 2020 were retrospectively analyzed, including 33 males aged 55.1±10.4 years, and 54 females aged 54.5±14.1 years. A total of 90 nodules were included, which were divided into a malignant tumor group (n=80) and a benign lesion group (n=10), and the malignant tumor group was subdivided into an invasive adenocarcinoma group (n=60) and a non-invasive adenocarcinoma group (n=20). The malignant probability and doubling time of each group were compared and its ability to predict the benign and malignant nodules and the invasion degree was analyzed.ResultsBetween the malignant tumor group and the benign lesion group, the malignant probability was significantly different, and the malignant probability could better distinguish malignant nodules and benign lesions (87.2%±9.1% vs. 28.8%±29.0%, P=0.000). The area under the curve (AUC) was 0.949. The maximum diameter of nodules in the benign lesion group was significantly longer than that in the malignant tumor group (1.270±0.481 cm vs. 0.990±0.361 cm, P=0.026); the doubling time of benign lesions was significantly longer than that of malignant nodules (1 083.600±258.180 d vs. 527.025±173.176 d, P=0.000), and the AUC was 0.975. The maximum diameter of the nodule in the invasive adenocarcinoma group was longer than that of the non-invasive adenocarcinoma group (1.350±0.355 cm vs. 0.863±0.271 cm, P=0.000), and there was no statistical difference in the probability of malignancy between the invasive adenocarcinoma group and the non-invasive adenocarcinoma group (89.7%±5.7% vs. 86.4%±9.9%, P=0.082). The AUC was 0.630. The doubling time of the invasive adenocarcinoma group was significantly shorter than that of the non-invasive adenocarcinoma group (392.200±138.050 d vs. 571.967±160.633 d, P=0.000), and the AUC was 0.829.ConclusionThe malignant probability and doubling time of lung nodules calculated by the artificial intelligence-assisted diagnosis and treatment system can be used in the assessment of the preoperative benign and malignant lung nodules and the infiltration degree.
In recent years, with the improvement of CT resolution, the reduction of radiation dose, the popularization of lung cancer screening and the enhancement of people's health awareness, the detection rate of lung nodules is higher and higher. Due to the close relationship between lung nodules and lung cancer, more and more attention has been paid to them. Although patients with early and middle stage lung cancer receive complete resection, all postoperative patients are at risk of recurrence and metastasis. Adjuvant or neoadjuvant therapy can improve the survival and reduce the recurrence and metastasis. Therefore, the multidisciplinary team, as the best model, provides a standardized and individualized plan for the diagnosis and treatment of lung nodules and lung cancer patients. However, in the clinical practice, the work efficiency of the multidisciplinary team is not high, and the participation rate of patients is low; therefore the multidisciplinary doctor model with thoracic surgeons as the mainstay is a reasonable alternative.
Objective To evaluate the predictive performance of the geometric characteristics, centerline (CL) of pulmonary nodules for prognosis in patients with surgically treatment in the National Lung Screening Trial (NLST). MethodsCT images of 178 patients who underwent surgical treatment and were diagnosed with non-small cell lung cancer (NSCLC) in the low-dose CT (LDCT) cohort from the NLST image database were selected, including 99 males and 79 females, with a median age of 64 (59, 68) years. CT images were processed using commercial software Mimics 21.0 to record the volume, surface area, CL and the area perpendicular to the centerline of pulmonary nodules. Receiver operating characteristic (ROC) curve was used to compare the predictive performance of LD, AD and CL on prognosis. Univariate Cox regression was used to explore the influencing factors for postoperative disease-free survival (DFS) and overall survival (OS), and meaningful independent variables were included in the multivariate Cox regression to construct the prediction model. ResultsThe area under the curve (AUC) of CL for postoperative recurrence and death were 0.650 and 0.719, better than LD (0.596, 0.623) and AD (0.600, 0.631). Multivariate Cox proportional risk regression analysis showed that pulmonary nodule volume (P=0.010), the maximum area perpendicular to the centerline (MApc) (P=0.028) and lymph node metastasis (P<0.001) were independent risk factors for DFS. Meanwhile, age (P=0.010), CL (P=0.043), lymph node metastasis (P<0.001), MApc (P=0.022) and the average area perpendicular to the centerline (AApc) (P=0.016) were independently associated with OS. ConclusionFor the postoperative outcomes of NSCLC patients in the LDCT cohort of the NLST, the CL of the pulmonary nodule prediction performance for prognosis is superior to the LD and AD, CL can effectively predict the risk stratification and prognosis of lung cancer, and spheroid tumors have a better prognosis.