IEEE J Biomed Health Inform. 2022 Oct;26(10):5004-5012. doi: 10.1109/JBHI.2022.3171663. Epub 2022 Oct 4.
Accurate classification of brain tumors is vital for detecting brain cancer in the Medical Internet of Things. Detecting brain cancer at its early stages is a tremendous medical problem, and many researchers have proposed various diagnostic systems; however, these systems still do not effectively detect brain cancer. To address this issue, we proposed an automatic diagnosing framework that will assist medical experts in diagnosing brain cancer and ensuring proper treatment. In developing the proposed integrated framework, we first integrated a Convolutional Neural Networks model to extract deep features from Magnetic resonance imaging. The extracted features are forwarded to a Long Short Term Memory model, which performs the final classification. Augmentation techniques were applied to increase the data size, thereby boosting the performance of our model. We used the hold-out Cross-validation technique for training and validating our method. In addition, we used various metrics to evaluate the proposed model. The results obtained from the experiments show that our model achieved higher performance than previous models. The proposed model is strongly recommended to be used to diagnose brain cancer in Medical Internet of Things healthcare systems due to its higher predictive outcomes.
在医疗物联网中，准确地对脑瘤进行分类对于发现脑癌至关重要。在早期发现脑癌是一个巨大的医学难题，许多研究人员已经提出了各种诊断系统；然而，这些系统仍然不能有效地检测出脑癌。为了解决这个问题，我们提出了一种自动诊断框架，将帮助医学专家诊断脑癌并确保进行适当的治疗。在开发所提出的集成框架时，我们首先集成了一个卷积神经网络模型，从磁共振成像中提取深度特征。提取的特征被转发到长短期记忆模型，该模型执行最终分类。应用了增强技术来增加数据量，从而提高了我们模型的性能。我们使用保留交叉验证技术来训练和验证我们的方法。此外，我们还使用了各种指标来评估所提出的模型。实验结果表明，我们的模型比以前的模型表现更好。由于具有更高的预测结果，强烈建议将所提出的模型用于医疗物联网医疗保健系统中的脑癌诊断。
