Machine learning-based CT radiomics features for the prediction of pulmonary metastasis in osteosarcoma.

OBJECTIVE: This study aims to build machine learning-based CT radiomic features to predict patients developing metastasis after osteosarcoma diagnosis. METHODS AND MATERIALS: This retrospective study has included 81 patients with a histopathological diagnosis of osteosarcoma. The entire dataset was divided randomly into training (60%) and test sets (40%). A data augmentation technique for the minority class was performed in the training set, along with feature's selection and model's training. The radiomic features were extracted from CT's image of the local osteosarcoma. Three frequently used machine learning models tried to predict patients with lung metastases (MT) and those without lung metastases (non-MT). According to the higher area under the curve (AUC), the best classifier was chosen and applied in the testing set with unseen data to provide an unbiased evaluation of the final model. RESULTS: The best classifier for predicting MT and non-MT groups used a Random Forest algorithm. The AUC and accuracy results of the test set were bulky (accuracy of 73% [ 95% coefficient interval (CI): 54%; 87%] and AUC of 0.79 [95% CI: 0.62; 0.96]). Features that fitted the model (radiomics signature) derived from Laplacian of Gaussian and wavelet filters. CONCLUSIONS: Machine learning-based CT radiomics approach can provide a non-invasive method with a fair predictive accuracy of the risk of developing pulmonary metastasis in osteosarcoma patients. ADVANCES IN KNOWLEDGE: Models based on CT radiomic analysis help assess the risk of developing pulmonary metastases in patients with osteosarcoma, allowing further studies for those with a worse prognosis.

Implicações do estadiamento do tumor de células gigantes do osso por ressonância nuclear magnética / Implications of magnetic resonance imaging in the radiographic staging of giant cell bone tumors for guidance of institutional research protocol

O tumor de células gigantes (TCG) do osso é uma lesão benigna, localmente agressiva, que compromete as extremidades dos ossos longos principalmente da articulação do joelho e o rádio distal. É um tumor que tem comportamento biológico imprevisível que não guarda relação com as características histológicas e, pela sua agressividade local pode acarretar em destruição óssea extensa e alta morbidade. A análise de imagens com boa qualidade técnica como as produzidas por ressonância nuclear magnética (RNM) aumenta a precisão diagnóstica do TCG, especialmente em relação ao comprometimento das partes moles. As lesões avaliadas em radiografias simples podem deixar de evidenciar o comprometimento tumoral extra ósseo, que está diretamente envolvido com maiores índices de recidiva, e necessita de tratamento cirúrgico com técnicas de resseção mais amplas. Devido ao alto custo, a RNM ainda é um exame de difícil acesso a população brasileira e a necessidade da sua realização em pacientes diagnosticados com TCG é reservada para o planejamento cirúrgico onde existe grande envolvimento extra-ósseo. O objetivo do presente estudo foi avaliar a precisão diagnóstica da invasão de partes moles do TCG pela radiologia convencional e por RNM em 21 pacientes portadores de TCG tratados no Instituto Nacional de Traumatologia e Ortopedia (INTO/MS). A partir da análise das radiografias, utilizando o sistema de estadiamento de Campanacci e Enneking, as lesões foram classificadas no estágio I em dois pacientes, no estágio II em nove pacientes e no estágio III em dez pacientes. A avaliação nas imagens radiográficas por RNM do estadiamento radiográfico mostrou que um paciente classificado no estágio I e todos classificados no estágio II tinham invasão de partes moles na RNM tratando-se, portanto, de lesões no estágio mais avançado (III). Outra característica das lesões que tinham invasão de partes moles não diagnosticadas no RX foi a presença de trabeculações em 9/10 pacientes. A partir dos resultados desse estudo baseado em evidências, propomos um fluxograma institucional para ser incluído na investigação préoperatória de pacientes com suspeita clínica de TCG. Acreditamos que nossos resultados possam servir para nortear o uso racional de recursos públicos destinados à investigação diagnóstica de lesões tumorais ósseas, em especial exames de alta complexidade e de alto custo como a RNM, priorizando a sua utilização para situações onde terá realmente impacto significativo no estadiamento da lesão e, consequentemente, na conduta terapêutica

The giant cell tumor (GCT) of bone is a benign lesion, locally aggressive, that affects the ends of long bones, mainly the knee joint and the distal radius. It is a tumor that has unpredictable biological behavior that is not related to the histological findings and, by its local aggressiveness can lead to extensive bone destruction and thus, high morbidity. The analysis with good technical quality image as magnetic resonance imaging (MRI) increases the diagnostic accuracy of the TCG, especially regarding the tumor extension to the soft tissues. The evaluation of the lesions on plain radiographs may fail to show tumor involvement outside to the bone, which is associated directly with higher recurrence rates and with the need of surgical treatment with wider resection techniques. Due to high costs, MRI is still an examination of difficult access to the population in general and the need for its realization in patients diagnosed with GCT is reserved for surgical planning in patients with extensive soft tissue and bone involvement. The aim of this study was to evaluate the diagnostic accuracy of tumor invasion of local soft tissues either by conventional radiology and magnetic resonance imaging, in 21 patients with GCT treated at the National Institute of Traumatology and Orthopedics (INTO/MS). The analysis of conventional radiographs, using the staging system of Campanacci and Enneking in two patients the lesions were classified as stage I, stage II in nine and in ten patients on stage III. The reassessment by MRI imaging of the radiographic staging showed that a patient classified on stage I and all patients classified on stage II had tumor involvement of adjacent soft tissues and were, therefore, lesions in the more advanced stage (III). Another feature of this subgroup of patients was the honeycomb appearance of the lesion seen in conventional radiographs in 9 out 10 patients. From the results of this study based on evidence, we propose an institutional flow chart to be used in the preoperative investigation of patients diagnosed with TCG. We believe that our results may serve to guide a rational use of public resources for diagnostic investigation of bone tumor lesions, especially tests of high complexity and high cost such as MRI, prioritizing their use to situations that will have a significant impact on the diagnosis and, consequently, in the therapeutic approach.

Magnetic resonance imaging aspects of giant-cell tumours of bone.

INTRODUCTION: This study aimed to describe the magnetic resonance imaging (MRI) features of giant-cell tumours of bone. METHODS: We analysed the clinical and MRI features of patients diagnosed with giant-cell tumours of bone confirmed by histopathology at our institution between 2010 and 2012. RESULTS: The peak incidence was between the second and third decades of life. There was no gender predominance. The most frequent locations were the knee and wrist. Pain and swelling were the prevailing symptoms. Fifty-one per cent of the patients were found to have associated secondary aneurysmal bone cysts on histopathology. On MRI, lesions demonstrated signal intensity equal to that of skeletal muscle on T1-weighted images and low signal intensity on T2-weighted images in 90% of cases. In gadolinium-enhanced T1-weighted images, 76.6% of cases demonstrated heterogeneous enhancement. We observed cystic components involving more than 50% of the lesion in 17 cases (56.6%). There was extra-osseous involvement in 13 cases (43.3%). CONCLUSION: MRI offers a valuable diagnostic tool for giant-cell tumours of bone. Contrast-enhanced MRI can distinguish between cystic and solid components of the tumour. MRI is also the imaging modality of choice for evaluation of soft-tissue involvement, offering a complete preoperative diagnosis.