Medical imaging technology upgrade! The radiation dose issue may be resolved

2024-01-10

X-ray transmission imaging/CT, as a non-invasive diagnostic method, is currently an important clinical detection method in the medical field. However, due to the ionization effect, X-rays can cause significant radiation damage to proteins, cells, etc. According to research reports from the Radiological Society, medical diagnosis of X-rays every year leads to a considerable number of cancer and leukemia patients. Therefore, it is crucial to reduce the dosage required for diagnosis. Since Roentgen discovered X-rays in 1895, there has been no fundamental change in the imaging methods. They all use direct projection onto surface detectors, which display a certain grayscale distribution by accumulating photons with object information. Therefore, the imaging efficiency of this method is very low, not only difficult to significantly reduce the required dose for imaging, but also limited by the size of the light source and the resolution of the detection equipment, Becoming the two major bottlenecks that constrain traditional imaging methods.

The concept of dose is related to the dosage of a medication, referring to the amount of medication taken by a patient at once. Later, it also referred to the amount of medication injected or otherwise administered to patients. But radiation is not a general "medicine", and it is difficult to use the word "give", but it depends on the amount of radiation energy received or absorbed. Therefore, the so-called "dose" only borrows this word, which is very different from its original meaning. Even in the field of ionizing radiation, there are many differences in the interpretation of the term dose. On the one hand, many scientists believe that this term cannot be considered a scientific term because it does not have a precise definition; On the other hand, it is necessary to use this so-called non terminological term extensively, because without it, it would be difficult to describe the topic we are facing. In view of this, it is necessary to provide one or several definitions for the term dose that are not defined below.

In view of the bottleneck problem of radiation dose, in 2018, Wu Ling'an, a researcher at the Institute of Physics of the Chinese Academy of Sciences/Key Laboratory of Photophysics of the National Research Center for Condensed Matter Physics, and Chen Liming cooperated to realize the mesa type X-ray ghost imaging for the first time by using a simple method of randomly modulating the light intensity. This indirect imaging method is based on the second-order correlation of the light field, and the imaging quality depends on the fluctuations of the detection signal rather than the intensity value. Based on this, the team completed single photon level ultra-low dose imaging, and the results received widespread attention after being published in Optica, which was reported by Science in the in-depth column. In a report by Science, experts in the same field gave high praise: "If applied to the field of medical imaging, this will be a revolutionary progress", while also expressing hope for the work: "improving the resolution and quality of imaging to meet the requirements of medical imaging". Based on the actual needs mentioned above, researcher Wu Ling'an from the Institute of Physics and current professor Chen Liming from Shanghai Jiao Tong University have once again collaborated to explore solutions to the imaging resolution bottleneck problem.

Pixel detector is a type of semiconductor detector that uses silicon as the detection material for particle track detection. The characteristic of pixel detectors is their excellent spatial resolution and rapid time response ability. In the ATLAS detector of the LHC in Europe, the sub detector near the beam is the pixel detector. In a 2cm × In a module with an area of 6cm, there are 47268 pixel units, and particles passing through any pixel will be recorded.

Recently, doctoral students He Yuhang and Zhang Aixin (co authored) in the research team used a self-developed Hadamard gold mask amplitude modulation board to achieve incoherent X-ray ghost imaging based on a true single pixel detector for the first time. Compared to the random modulation scheme, this method utilizes the orthogonal completeness property of the Hadamard matrix, so it can reconstruct better images even under sparse sampling. On this basis, the original algorithm was upgraded by introducing compressive sensing and convolutional neural networks, ultimately utilizing 37 μ The X light source with a size of m obtained 10 at a sampling rate of only 18.75% μ The imaging results of m resolution have achieved super-resolution imaging that breaks through the limitation of source size, which is sufficient for direct judgment of cancerous tissue and meets the resolution requirements of clinical medical fine imaging. Under the framework of computing ghost imaging, high-performance algorithms and the fine structure of modulation plates ensure better image quality under super-resolution, while lower sampling rates mean shorter exposure time and lower dose. Therefore, this technology is expected to further reduce dose. The entire experimental layout is simple, convenient to use, and the application of single pixel detectors can greatly reduce costs. On the other hand, the application of this method greatly reduces the requirements for spatial coherence and intensity of radiation sources, which can greatly promote the practical process of X-ray ghost imaging.

Quantum correlation imaging technology successfully applied in the X-ray band