Low-Dose Computed Tomography (LDCT) is the most common imaging modality for lung cancer diagnosis. The presence of nodules in the scans does not necessarily portend lung cancer, as there is an intricate relationship between nodule characteristics and lung cancer. Therefore, benign-malignant pulmonary nodule classification at early detection is a crucial step to improve diagnosis and prolong patient survival. The aim of this study is to propose a method for predicting nodule malignancy based on deep abstract features. Methods: To efficiently capture both intra-nodule heterogeneities and contextual information of the pulmonary nodules, a dual pathway model was developed to integrate the intra-nodule characteristics with contextual attributes. The proposed approach was implemented with both supervised and unsupervised learning schemes. A random forest model was added as a second component on top of the networks to generate the classification results. The discrimination power of the model was evaluated by calculating the Area Under the Receiver Operating Characteristic Curve (AUROC) metric. Results: Experiments on 1297 manually segmented nodules show that the integration of context and target supervised deep features have a great potential for accurate prediction, resulting in a discrimination power of 0.936 in terms of AUROC, which outperformed the classification performance of the Kaggle 2017 challenge winner. Conclusion:Empirical results demonstrate that integrating nodule target and context images into a unified network improves the discrimination power, outperforming the conventional single pathway convolutional neural networks.

ObjectivesBoth radiomics and deep learning methods have shown great promise in predicting lesion malignancy in various image-based oncology studies. However, it is still unclear which method to choose for a specific clinical problem given the access to the same amount of training data. In this study, we try to compare the performance of a series of carefully selected conventional radiomics methods, end-to-end deep learning models, and deep-feature based radiomics pipelines for pulmonary nodule malignancy prediction on an open database that consists of 1297 manually delineated lung nodules.MethodsConventional radiomics analysis was conducted by extracting standard handcrafted features from target nodule images. Several end-to-end deep classifier networks, including VGG, ResNet, DenseNet, and EfficientNet were employed to identify lung nodule malignancy as well. In addition to the baseline implementations, we also investigated the importance of feature selection and class balancing, as well as separating the features learned in the nodule target region and the background/context region. By pooling the radiomics and deep features together in a hybrid feature set, we investigated the compatibility of these two sets with respect to malignancy prediction.ResultsThe best baseline conventional radiomics model, deep learning model, and deep-feature based radiomics model achieved AUROC values (mean ± standard deviations) of 0.792 ± 0.025, 0.801 ± 0.018, and 0.817 ± 0.032, respectively through 5-fold cross-validation analyses. However, after trying out several optimization techniques, such as feature selection and data balancing, as well as adding context features, the corresponding best radiomics, end-to-end deep learning, and deep-feature based models achieved AUROC values of 0.921 ± 0.010, 0.824 ± 0.021, and 0.936 ± 0.011, respectively. We achieved the best prediction accuracy from the hybrid feature set (AUROC: 0.938 ± 0.010).ConclusionThe end-to-end deep-learning model outperforms conventional radiomics out of the box without much fine-tuning. On the other hand, fine-tuning the models lead to significant improvements in the prediction performance where the conventional and deep-feature based radiomics models achieved comparable results. The hybrid radiomics method seems to be the most promising model for lung nodule malignancy prediction in this comparative study.

Computed tomography (CT) was introduced to medicine in the early 1970s, which brought slice imaging into wide use for the first time. Today, CT is an essential part of radiological diagnostics, and is used for a wide range of clinical applications. One downside of CT imaging is the health risks related to the ionizing radiation. In the 80s it was believed that CT would soon be replaced completely by MRI due, in part, to the ionizing radiation required in CT. A further downside are the health risks related to the use of iodine-based contrast media. Both radiation-and contrast media dose have a trade-off with image quality. However, many technical advances have been made, and progress is still ongoing, to improve and broaden the applications of CT. Such advances necessitate a re-evaluation of imaging protocols and continued optimization of radiation dose, contrast media dose and imaging quality. This is the subject of this PhD project.Study I: The aim of this study was to evaluate the potential of low-kV dual-source (DS) and dual-energy (DE) to reduce CM-doses while maintaining soft-tissue and iodine CNR in phantoms of varying size, and to quantify the corresponding radiation dose increases. It was found that low-kV dual-source imaging could be used to reduce CM doses by 44-53% with maintained iodine-, soft tissue-and other materials CNR in a wide range of abdominal sizes, to the cost of about 20-100% increased radiation dose, depending on size. The dual-energy technique allowed a reduction of CM dose by 20% at similar radiation dose as the standard 120 kV protocol.Study II: The aim of this study was to implement and evaluate a scanning regimen, based on the results from Study I, to reduce CM-doses for patients believed to be at risk of CIN. It was concluded that the protocols from Study I could be used to reduce CM doses by 40-50%, depending on patient size, with maintained CNR in patients with a BMI-range of 15-36 kg/m 2 . The size-specific dose estimates increased by 70%.Study III: The aim of this study was to compare the outcome in image noise and radiation dose in the subsequent CT scan following a single anterior-posterior (AP) vs a combined lateral plus AP (LAT+AP) localizer when using automatic tube-current modulation (ATCM). The results suggested that using LAT+AP localizer yields more consistent noise and radiation dose than a single AP. The effect was small, except for a subgroup of females with laterally protruding breast tissue, which may have been overexposed by about 57% in the thorax region.Study IV: The aim of this study was to evaluate if standard-dose CT can be replaced with low-dose CT for characterization of non-specific findings bone in scintigraphy. Based on these results, sub-mSv CT seems feasible for morphological characterization of skeletal changes in areas with increased tracer uptake on bone scintigraphy, although a larger study is needed.