Reading experience of "Deep Learning and Medical Image Processing" 1

kkompp

Reading experience of "Deep Learning and Medical Image Processing" 1 [Copy link]

First of all, I would like to thank the forum for giving me the opportunity to participate in the reading of "Deep Learning and Medical Image Processing". I sincerely thank the forum.

First impression

The cover of the book looks like this

The book was published by People's Posts and Telecommunications Press and written by five experts. The preface was written by President Wang of Beijing Tiantan Hospital affiliated to Capital Medical University, highlighting the professionalism of the book. Three of the five authors are computer scientists and two are medical experts, ensuring that the content of the book is closely related to reality.

The programming languages used in the book are mainly conventional Python deep learning, such as Python, Numpy, TensorFlow, Matplotlib, Sciki, etc. Pydicom, elasticdeform and simpletk seem to be relatively rare in previous deep learning applications. Maybe I am not well-informed, so please forgive me for any shortcomings.

2. Contents of this book

This book is a professional book on the application of artificial intelligence in the field of medical imaging. As we all know, doctors in modern hospitals mainly rely on the detection indicators of various instruments to diagnose diseases, such as electrocardiograms, x-rays, CT, MRI and other images that we check in hospitals. Doctors make diagnostic analyses based on the characteristics of the images and the differences from normal images. This constitutes the main content of current hospital diagnosis cases. This book helps doctors make diagnoses by learning and analyzing medical images through deep learning, deriving pathological features, and giving diagnostic results.

The book is divided into 10 chapters:

The first chapter mainly introduces the application of artificial intelligence in medical images. It is an overview and you only need to understand it basically.

Chapter 2 is an introduction to medical data, including X-ray imaging, MRI imaging, ultrasound imaging, electrocardiogram and other images generated by various instruments used in hospitals, which form the basis of deep learning; these images are represented in different formats, which are different from the common jpg images we see, but are medical-specific image formats.

Chapter 3 is about data annotation based on Chapter 2. This is done by a special software, 3D Slicer, which performs segmentation and reconstruction on the medical data in various formats mentioned in Chapter 2, which can be understood as data preprocessing.

Chapter 4 is medical digital image processing, which is pure image processing that is independent of the medical profession, such as interpolation, data normalization, etc. Data enhancement uses TensorFlow methods.

From Chapter 5 to Chapter 10, they are about medical image processing and model learning optimization, such as semantic segmentation, key point detection, medical image registration, model optimization, and transfer learning. This part of the content requires hard work and in-depth research, and this is the value of this book.

Chapter 3: Application of Artificial Intelligence in Medicine

The first chapter is an overview, mainly about the introduction of artificial intelligence and machine learning. Machine learning is an interdisciplinary subject, in fact, it is an interdisciplinary mathematics, involving probability, statistics, convex optimization and other subjects. Due to its mathematical properties, it constitutes a certain entry threshold. The concept of agent is also mentioned here. There have been many achievements in the research of agent in academia. Every machine learning host is an agent.

The application of artificial intelligence in the medical field is more reflected in assisting medical diagnosis. As for clinical treatment and disease prediction, due to the professional nature of monitoring images, I personally think it is still immature.

The applications of artificial intelligence in the medical field include: medical image acquisition and reconstruction, image transformation, lesion area detection and segmentation, and intelligent diagnosis. These are what deep learning is best at, but the application scenario here is medical image analysis.

This book lists references after each chapter, which is also a practice favored by academic scientists. We can see that there are 43 references listed in the first chapter, most of which are published in foreign literature, including some of the author's own articles. This shows the accumulation of the author's professional knowledge.

Four conclusions

Another interesting thing about this book is that the images are in color and the table of contents is in blue, which shows the seriousness and responsibility in the typesetting and also displays a certain professionalism. It is very interesting.

Next, you need to analyze the contents of the book carefully, study it in depth, and strive to enter the door of medical image processing.