Diagnostic radiology has undergone three main evolutionary stages since Roentgen’s initial discovery of x-rays in 1895.
The period from 1895 to 1918 may be termed the experimentation period when the application of x-rays and their production were explored. However, this period culminated in practical clinical trials of the use of X-rays during the First World War when radiology was tested on the battlefield. The large number of casualties suffering from gunshot wounds provided an ideal testing ground for the detection of high contrast metallic objects located inside patients. An X-ray examination enabled medical staff to determine which casualties had the highest probability of survival prior to any surgical intervention. This facilitated the best use of limited surgical capacity. The application of X-ray imaging on the battlefield also led to the development of role of the technical or radiographic assistant with the interpretation of images provided by medical practitioners (radiologists).
When World War 1 broke out in 1914, Marie Curie immediately saw the potential of X-rays to save lives on the battlefield. Together with her daughter, she helped to establish a network of battlefield radiological stations that incorporated some 200 mobile field units, known as petites Curies. Consequently, the, setting up and operating of such facilities in order to produce acceptable images required a degree of technical knowledge and capability.
From 1918 to 1972 diagnostic radiology became an accepted and useful diagnostic technique within healthcare and large departments were established in many major hospitals as well as in peripheral smaller local “cottage hospital”. Indeed, up until the 1970’s many cottage hospitals still employed quite ancient X-ray equipment that had stud settings for kV selection. Since surgical intervention represented a primary clinical intervention, radiology provided a useful input. During this period, the necessary staffing and operating infrastructure for diagnostic radiology was developed as well as training programmes. The application of fluoroscopy and tomography also became common place to support conventional radiography.
1972 may be taken to be the beginning of the transformation of diagnostic radiology from a clinically proven and mainly film based analogue process into its present status as a key element of today’s modern healthcare system. Indeed, diagnostic radiology led the way in transforming healthcare into a technology driven activity. The point of transition into the modern era is demonstrated by four presentations made at the 1972 RSNA Conference in Chicago that in retrospect indicated the emergence of a new order (1). These were:
Course no 101: The use of computers in diagnostic radiology presented by Howard J Barnard
The Symposium; New techniques in Radiology: Computerized Axial Tomography presented by James Ambrose and Godfrey Hounsfield
The section on Radiation Therapy and Radiobiology: Patient information retrieval using a medium sized computer presented by Anthony Chung-Bin, Thomas Wachtor et al
Course no 501: Decision making studies and computers presented by Lee Lusted
These four topics represent key pillars of modern radiological practice and this period also coincided with the development of the first commercially successful minicomputer produced by Digital Equipment Corporation (DEC). This technological development provided the basis for localized data processing and image reconstruction functionality at a reasonable cost for the early CT scanners.
From 1972 onwards there was a continuous integration of technological developments led by the introduction of CT. These have provided the focus for the development of the modern radiological practices we see today. CT was a truly revolutionary digital imaging system and introduced the concept of improved contrast detectability into X-ray imaging. It demonstrated that spatial resolution, provided by screen-film systems, could be sacrificed for clinical gain through improved contrast detectability. This was followed shortly afterwards by the development of digital subtraction angiography, which also enhanced clinical performance through improved contrast detectability. The improvement arose from the removal of anatomical structural “noise” that was known to influence the detectability of abnormalities in clinical images. Initially this technique employed digitization of the anologue video signal from the fluoroscopic TV camera. The subtraction of images taken both prior to and post-injection of contrast agents produced images of a contrast enhanced vascular pattern.
These developments had let the digital cat out of the bag and fuelled ongoing and rapid developments driven by large scale investments by industry in micro-chip technology that led to the IT capabilities we take for granted today. Significant investments were also made in new X-ray imaging detectors, new imaging modalities and rapid increases in data storage capabilities. For example, as recently as the early 1980’s the goal was to produce 1 gigabyte optical disk ROM devices for data storage! In the intervening period virtually unlimited storage capacity has now been created through vast amounts of cloud data storage.
This evolutionary process was supported by the development and integration of data management functions into RIS and PACS, which modernised work practices, previously based on analogue (paper and film) information sources. The days of a hospital having to store historical X-ray film images have long gone. These concepts evolved from the development of the World Wide Web in 1989 as a tool to meet demand for automated information sharing between scientists and universities around the world.
The World-wide web became the internet, which provided the operational platform for broadband telecommunications, which now plays such a crucial role in modern society. However, in this period major developments also occurred in every facet of the imaging system from the X-ray tube, which supported the rapid production of high volumes of imaging data. Equally, developments continued in the fields of data management, communications and storage through to the image display. Many of these developments were equally applicable to the mass consumer market.
The relentless progress that occurred in both the technological development as well as the significant growth in the application of diagnostic radiology is reflected in the expansion in digital information storage since 1986 (2). In 1986 the global information storage capacity in optimally compressed bytes was 2.6 exabytes of analogue storage and 0.02 exabytes of digital. (1 exabyte equals 1018 bytes). By 2002, which is classed as the beginning of the digital age, digital storage equalled the analogue storage capacity. By 2007, 94 % of storage was digital and equal to 280 exabytes with 4% analogue of 19 exabytes. In 2013 the volume of healthcare data alone was indicated to be 152 exabytes with a projected growth rate of 36%. Thus by 2020 it was indicated to be 2,314 exabytes (3), pre-covid-19 with single patients generating nearly 80 megabytes of data each year in imaging and EMR (Electronic Medical Records) (4). At present the monetary value of this storage is estimated to be $4.17 billion (5). In fact, healthcare is indicated to be generating approximately 30% of all the world’s data-volume, with continued growth anticipated, given the rudimentary healthcare facilities at present available to much of the world’s population.
Thus, from 1972 to 2012 developments were predominantly driven by a desire to improve and modernise the radiology process itself. However, from the early 2000’s a shift towards outcome driven developments began. Already by this time it was noted that medical imaging was generating more data than could be usefully assessed and analysed by human observers in reasonable time scales (6). A hospital’s data storage needs were now largely dictated by the imaging department. Thus, greater effort was needed in developing mechanisms for more effectively assessing the increasing quantities of medical image data as well as the capability to integrate different sources of clinical data into patient care. The human element was now a significant limitation for fast, efficient and consistent reporting in diagnostic radiology. Consequently, we appear to be entering into a new era driven by the application of mechanisms such as AI and Machine Learning techniques to enable humans to effectively employ such enormous volumes of data in clinical practice.