Does aggressive management of solitary pulmonary nodules pay off?

Classification of Pulmonary Nodules in 2-[18F]FDG PET/CT Images with a 3D Convolutional Neural Network

Purpose

2-[18F]FDG PET/CT plays an important role in the management of pulmonary nodules. Convolutional neural networks (CNNs) automatically learn features from images and have the potential to improve the discrimination between malignant and benign pulmonary nodules. The purpose of this study was to develop and validate a CNN model for classification of pulmonary nodules from 2-[18F]FDG PET images.

Methods

One hundred thirteen participants were retrospectively selected. One nodule per participant. The 2-[18F]FDG PET images were preprocessed and annotated with the reference standard. The deep learning experiment entailed random data splitting in five sets. A test set was held out for evaluation of the final model. Four-fold cross-validation was performed from the remaining sets for training and evaluating a set of candidate models and for selecting the final model. Models of three types of 3D CNNs architectures were trained from random weight initialization (Stacked 3D CNN, VGG-like and Inception-v2-like models) both in original and augmented datasets. Transfer learning, from ImageNet with ResNet-50, was also used.

Results

The final model (Stacked 3D CNN model) obtained an area under the ROC curve of 0.8385 (95% CI: 0.6455-1.0000) in the test set. The model had a sensibility of 80.00%, a specificity of 69.23% and an accuracy of 73.91%, in the test set, for an optimised decision threshold that assigns a higher cost to false negatives.

Conclusion

A 3D CNN model was effective at distinguishing benign from malignant pulmonary nodules in 2-[18F]FDG PET images.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13139-023-00821-6.

Radiomics and Artificial Intelligence Can Predict Malignancy of Solitary Pulmonary Nodules in the Elderly

Solitary pulmonary nodules (SPNs) are a diagnostic and therapeutic challenge for thoracic surgeons. Although such lesions are usually benign, the risk of malignancy remains significant, particularly in elderly patients, who represent a large segment of the affected population. Surgical treatment in this subset, which usually presents several comorbidities, requires careful evaluation, especially when pre-operative biopsy is not feasible and comorbidities may jeopardize the outcome. Radiomics and artificial intelligence (AI) are progressively being applied in predicting malignancy in suspicious nodules and assisting the decision-making process. In this study, we analyzed features of the radiomic images of 71 patients with SPN aged more than 75 years (median 79, IQR 76-81) who had undergone upfront pulmonary resection based on CT and PET-CT findings. Three different machine learning algorithms were applied-functional tree, Rep Tree and J48. Histology was malignant in 64.8% of nodules and the best predictive value was achieved by the J48 model (AUC 0.9). The use of AI analysis of radiomic features may be applied to the decision-making process in elderly frail patients with suspicious SPNs to minimize the false positive rate and reduce the incidence of unnecessary surgery.

Pilot study using virtual 4-D tracking electromagnetic navigation bronchoscopy in the diagnosis of pulmonary nodules: a single center prospective study

Background

Electromagnetic navigation bronchoscopy (ENB) is a navigation technology intended to improve the diagnostic yield of pulmonary nodules. However, nodule displacement due to respiratory motion may compromise the accuracy of the navigation guidance. The Veran SPiNDrive ENB system employs respiratory-gating (4D-tracking) to compensate for this motion. The aim of the present study was to evaluate the diagnostic performance and safety of the Veran SPiNDrive system for biopsy of pulmonary nodules.

Methods

Adult patients with pulmonary nodules of ≥1 cm were enrolled at a single center. Both conventional bronchoscopy and 4D-tracking ENB were performed in one procedure session under general anesthesia, with the procedure order being randomly assigned. Radial probe endobronchial ultrasound and fluoroscopy were used in both groups. The diagnostic performance, safety, total procedure time, and total fluoroscopy time of the ENB phase were compared to the corresponding conventional bronchoscopy phase.

Results

The study was terminated due to poor accrual; a total of eleven patients were enrolled. The mean size of pulmonary nodules was 2.1 cm. The sensitivity for malignancy was 67% (6/9) and 56% (5/9) with conventional bronchoscopy and with 4D-tracking ENB, respectively. Two cases developed minor bleeding after conventional bronchoscopy, while no complications were observed after 4D-tracking ENB. The mean procedure time was 16.1 and 21.7 min (P=0.090), and the mean duration time for fluoroscopy use was 77 and 44 sec (P=0.056) for the conventional bronchoscopy and the 4D-tracking ENB phases, respectively.

Conclusions

The diagnostic performance of the Veran SPiNDrive 4D-tracking ENB did not exceed that of conventional bronchoscopy for pulmonary nodules. No complications were seen during 4D-tracking ENB. A study with a larger number of participants is required for further assessment.

Lo, the ever confounding nipple shadow!

The discovery of a solitary pulmonary nodule (SPN) on chest imaging can be alarming for both the clinician and the patient. In the absence of a uniform guideline, managing SPN is nothing short of challenging for primary care physicians (PCPs). We present a case here of a patient presenting with prolonged cough who also displayed unilateral SPN on her chest radiograph. Through further examination, this presence was later shown to be a nipple shadow simulating SPN, and the patient was spared unnecessary testing and psychological distress.

Detecting small pulmonary nodules with spiral ultrashort echo time sequences in 1.5 T MRI

OBJECTIVE

This study investigated ultrashort echo time (UTE) sequences in 1.5 T magnetic resonance imaging (MRI) for small lung nodule detection.

MATERIALS AND METHODS

A total of 120 patients with 165 small lung nodules before video-associated thoracoscopic resection were enrolled. MRI sequences included conventional volumetric interpolated breath-hold examination (VIBE, scan time 16 s), spiral UTE (TE 0.05 ms) with free-breathing (scan time 3.5-5 min), and breath-hold sequences (scan time 20 s). Chest CT provided a standard reference for nodule size and morphology. Nodule detection sensitivity was evaluated on a lobe-by-lobe basis.

RESULTS

The nodule detection rate was significantly higher in spiral UTE free-breathing (> 78%, p < 0.05) and breath-hold sequences (> 75%, p < 0.05) compared with conventional VIBE (> 55%), reaching 100% when nodule size was > 16 mm, and reaching 95% when nodules were in solid morphology, regardless of size. The inter-sequence reliability between free-breathing and breath-hold spiral UTE was good (κ > 0.80). Inter-reader agreement was also high (κ > 0.77) for spiral UTE sequences. Nodule size measurements were consistent between CT and spiral UTE MRI, with a minimal bias up to 0.2 mm.

DISCUSSION

Spiral UTE sequences detect small lung nodules that warrant surgery, offers realistic scan times for clinical work, and could be implemented as part of routine lung MRI.

Analysis of solitary pulmonary nodule after surgical resection in patients with 18F-FDG positron emission tomography integrated computed tomography in the preoperative work-up

INTRODUCTION

To analyse clinicopathological characteristics of patients operated for pulmonary solitary nodule (PSN) and 18F-FDG integrated PET-CT scan after surgical resection.

METHODOLOGY

Retrospective study on a prospective database of patients operated from January 2007 to October 2017 for PSN without preoperative diagnosis. Dependent variable was anatomopathological result (benign vs malignant) of PSN. Variables of the study were: age, sex, PET-CT uptake, SUVmax, smoking history, COPD, previous history of malignant disease, tumoral location, and tumour size on CT-scan.

RESULTS

A total of 305 patients were included in this study, 225 (73.8%) men, 80 (26.2%) women, mean age = 63.9 (range 29-86 years), mean size PSN = 1.68 (s.d. .65 cm), benign = 46 (15.1%), malignant = 258 (84.6%), type of resection: pulmonary wedge = 151 (49.5%), lobectomy = 141 (46.2%), segmentectomy = 12 (3.9%), exploratory intervention = 1 (0.3%). Postoperative mortality was 1.9%. COPD = 50.8% cases, previous cancer disease = 172 cases (56.4%), smoking history = 250 cases (82.0%), positive PET = 280 cases (91.8%), PSN in upper pulmonary fields = 204 cases (66.9%), median SUVmax = 3.4 (range 0-20.7). Backward stepwise binary logistic regression model showed that age, SUVmax, previous malignant disease and female sex were independent risk factors with statistical significance (p < .05). Sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy were 94.6%, 23.4%, 87.1%, 44.0%, and 83.6% respectively. There were 14 false negative cases (4.6%) and 36 false positive cases (11.8%).

CONCLUSIONS

Age, SUVmax, previous malignant disease, and female sex were independent risk factors in our study. Each case should be individually evaluated in a multidisciplinary committee, and the patient's preferences or concerns should be kept in mind in decision-making. Surgical resection of PSN is not exempt from morbidity and mortality, even in sublobar or pulmonary wedge resection.

Too much or too little medicine? Overdiagnosis, underdiagnosis, overtreatment and undertreatment in respiratory diseases

At a time when technical possibilities for medical investigations are plentiful and ever expanding, there is growing awareness that more is not always better and that “too much medicine” may be harmful. In recent years, the global Choosing Wisely educational campaign has aimed to bring attention to unnecessary healthcare. Unnecessary healthcare includes overtesting, overdiagnosis and overtreatment [1]. The March issue of Breathe aims to challenge us to rethink our clinical practice, to reflect on the evidence, and to identify potential cognitive biases that might influence us to provide “too much” or “too little” medicine [2].

The March issue of Breathe focuses on overdiagnosis, underdiagnosis, overtreatment and undertreatment in respiratory diseaseshttp://ow.ly/63OW30ntCeu