CARLOS SHIMIZU
Índice h a partir de 2011
5
Projetos de Pesquisa
Unidades Organizacionais
Instituto do Câncer do Estado de São Paulo, Hospital das Clínicas, Faculdade de Medicina - Médico
LIM/24 - Laboratório de Oncologia Experimental, Hospital das Clínicas, Faculdade de Medicina
LIM/24 - Laboratório de Oncologia Experimental, Hospital das Clínicas, Faculdade de Medicina
6 resultados
Resultados de Busca
Agora exibindo 1 - 6 de 6
bookPart Procedimentos guiados por ressonância magnética(2022) HSIEH, Su Jin Kim; SHIMIZU, Carlos- Pre-treatment MRI tumor features and post-treatment mammographic findings: may they contribute to refining the prediction of pathologic complete response in post-neoadjuvant breast cancer patients with radiologic complete response on MRI?(2022) THOMPSON, Bruna M.; CHALA, Luciano F.; SHIMIZU, Carlos; MANO, Max S.; FILASSI, Jose R.; GEYER, Felipe C.; TORRES, Ulysses S.; MELLO, Giselle Guedes Netto de; LEITE, Claudia da CostaPurpose Radiologic complete response (rCR) in breast cancer patients after neoadjuvant chemotherapy (NAC) does not necessarily correlate with pathologic complete response (pCR), a marker traditionally associated with better outcomes. We sought to verify if data extracted from two important steps of the imaging workup (tumor features at pre-treatment MRI and post-treatment mammographic findings) might assist in refining the prediction of pCR in post-NAC patients showing rCR. Methods A total of 115 post-NAC women with rCR on MRI (2010-2016) were retrospectively assessed. Pre-treatment MRI (lesion morphology, size, and distribution) and post-treatment mammographic findings (calcification, asymmetry, mass, architectural distortion) were assessed, as well as clinical and molecular variables. Bivariate and multivariate analyses evaluated correlation between such variables and pCR. Post-NAC mammographic findings and their correlation with ductal in situ carcinoma (DCIS) were evaluated using Pearson's correlation. Results Tumor distribution at pre-treatment MRI was the only significant predictive imaging feature on multivariate analysis, with multicentric lesions having lower odds of pCR (p = 0.035). There was no significant association between tumor size and morphology with pCR. Mammographic residual calcifications were associated with DCIS (p = 0.009). The receptor subtype remained as a significant predictor, with HR-HER2 + and triple-negative status demonstrating higher odds of pCR on multivariate analyses. Conclusions Multicentric lesions on pre-NAC MRI were associated with a lower chance of pCR in post-NAC rCR patients. The receptor subtype remained a reliable predictor of pCR. Residual mammographic calcifications correlated with higher odds of malignancy, making the correlation between mammography and MRI essential for surgical planning.
bookPart Atribuições dos residentes/imaginologistas mamários(2022) HSIEH, Su Jin Kim; SHIMIZU, CarlosbookPart Ressonância magnética das mamas(2022) CHALA, Luciano Fernandes; SHIMIZU, CarlosbookPart Correlação entre os métodos de imagem e os aspectos clínicos e dicas práticas(2022) SHIMIZU, Carlos; BARROS, Nestor de- Breast Cancer Diagnosis in Two-View Mammography Using End-to-End Trained EfficientNet-Based Convolutional Network(2022) PETRINI, Daniel G. P.; SHIMIZU, Carlos; ROELA, Rosimeire A.; VALENTE, Gabriel Vansuita; FOLGUEIRA, Maria Aparecida Azevedo Koike; KIM, Hae YongSome recent studies have described deep convolutional neural networks to diagnose breast cancer in mammograms with similar or even superior performance to that of human experts. One of the best techniques does two transfer learnings: the first uses a model trained on natural images to create a ""patch classifier"" that categorizes small subimages; the second uses the patch classifier to scan the whole mammogram and create the ""single-view whole-image classifier"". We propose to make a third transfer learning to obtain a ""two-view classifier"" to use the two mammographic views: bilateral craniocaudal and mediolateral oblique. We use EfficientNet as the basis of our model. We ""end-to-end"" train the entire system using CBIS-DDSM dataset. To ensure statistical robustness, we test our system twice using: (a) 5-fold cross validation; and (b) the original training/test division of the dataset. Our technique reached an AUC of 0.9344 using 5-fold cross validation (accuracy, sensitivity and specificity are 85.13% at the equal error rate point of ROC). Using the original dataset division, our technique achieved an AUC of 0.8483, as far as we know the highest reported AUC for this problem, although the subtle differences in the testing conditions of each work do not allow for an accurate comparison. The inference code and model are available at https://github.com/dpetrini/two-views-classifier