Radiology for Residents

The Beatles, the Nobel Prize, and CT Scanning of the Chest


Lawrence R. Goodman, MD

On June 6, 1962, the Beatles (Fig. 1A) had their first recording session with Electrical Musical Industries, Ltd (EMI). Their meteoric success changed the history of modern radiology and medicine forever. The money generated by record sales enabled the EMI basic science researchers to thrive in a cash-rich environment,1–3 including the research of Dr. Godfrey Hounsfield, an electri- cal and computer engineer. He had spent years exploring whether back projection methods of producing an image could use differential X-ray attenuation values.

‘‘YESTERDAY. SEEMED SO FAR AWAY’’


In 1967, the first experimental computer axial tomography (CAT) scan was constructed on an old lathe, using americium as a gamma ray source. The scan of a mouse took 9 days to complete. It required 2.5 hours of main frame computer time to reconstruct, but produced a recognizable image. Four years later, in October of 1971, the first head scan of a living patient was performed using an EMI ‘‘Mark I’’ scanner. The equipment used a translate-rotate gantry (step and shoot), an 80 x 80 matrix yielding a spatial resolution of 0.5 cm, and required a water bag for stabilization and normalization of the head. Reconstruction took all night but produced a recognizable image of a brain tumor. ‘‘My God, it does work!’’ exclaimed Hounsfield. The first EMI production model required 4 minutes per slice and 7 minutes per reconstruction. The first description of a CAT scan in the radiology literature by Hounsfield and colleagues was in the British Journal of Medicine in 1973.

Hounsfield apparently was unaware of prior work. In the 1960s, Dr Allan Cormack of South Africa, and later Tufts’ University, a particle physicist, and Dr William Oldendorf, a Colorado neurologist, independently showed that multiple measures of radiograph attenuation around a target enabled one to compute an image of that target. Unfortunately, without more powerful computers, there was little practical application of this concept.
Although Cormack (PhD) and Hounsfield (no formal degree) never met, and neither had a medical background or interest in medicine, both received the Nobel Prize in Physics and Medicine, in 1979, for the CAT scan, the ‘‘greatest advance in radiologic medicine since the discovery of the X-ray.’’ Cormack was cited for his math analysis that led to the CAT scan and Hounsfield for its practical development (Figs.1B and 2).10 As CAT scans became more sophisticated, new areas of investigation opened up in radiology. Many opened up new approaches to surgery, and new understandings of medical conditions emerged.

‘‘WITH A LITTLE HELP FROM (OUR) FRIENDS’’


EMI estimated it needed to sell 25 CAT scanners worldwide to make it a commercially viable product. Others were more optimistic. Over the next few years, 18 companies, large and small, competed for the growing scanner market. Over the first decade, many companies disappeared, leaving several major manufacturers as survivors.

Stacks Image 93


Fig. 1. (A) The Beatles. (B) The Nobel Prize Medallier. (C) Early chest CT scan from 1973. (Fig. 1C. From Sheedy PF 2nd, Stephens DH, Hattery RR, et al. Computed tomography of the body: initial clinical trial with the EMI proto- type. AJR Am J Roentgenol 1976;127:23–51; with permission.)


Gradually, computed axial tomography (CAT scanning) morphed into computed tomography (CT scanning). The first body CT scanner, which required no water bag, was developed at Georgetown Hospital by Dr Robert Ledley, a dentist (Fig. 3). The automatic computerized transverse axial (ACTA) scanner was commercialized by Pfizer, with 30 photomultiplier tubes and a 256 x 256 matrix. The race to perform faster and better CT scans was on! Improvements in the translate-rotate scans by several manufacturers brought scan time down to 2 minutes per image (Fig. 4). By 1974, second- generation scanners from EMI, Ohio Nuclear, and Siemens brought scan time down to 1 minute, and then to18 seconds (Figs. 5 and 6). A 320 x 320 matrix replaced an 80 x 80 matrix.

In 1974, General Electric (GE) abandoned the translate-rotate approach and proposed a fan beam, which rotated around the patient (rotate- rotate) in synchronicity with a small curvilinear detector (300–700 elements) (see Fig. 4). This was a small scanner prototype large enough to image the breast and, eventually, the head. By 1976, they produced a third-generation body scanner capable of producing 10 mm axial images with 9.5-second gantry rotation per image. Within a few years, GE and Siemens were the dominant producers of third-generation scanners.
Fourth generation scanners, produced by Technicare (Johnson & Johnson), mounted a thousand stationary detectors around the gantry with a rotating radiograph tube. Technical problems eventually defeated this technique. Imaging changed dramatically in 1989, when the first helical or spiral scanners were produced by Siemens. The tube rotated continuously as the patient moved through the gantry. Elscint produced a two-detector scanner and the multi-detector race was on.

‘‘YOU REALLY GOT A HOLD OF ME’’


The life of the radiologist, and every physician, hospital, and patient, was never the same again. Although the original mages were quite crude, the machinery was expensive, and validating studies were lacking, the CT scans were an instantaneous success.1 It required only a brief look to realize that the CT scans were special and would replace many conventional techniques. By 1979, 6 years after its clinical introduction, 1,300 CT scanners were in use in the United States. By 1980, 3 million CT scans were performed and, by the year 2000, 62 million CT scans were performed annually. With each generation of scanner, new applications arose, much of it made possible by the rapid increase in computer power and the markedly decreasing cost of memory.

The first published images of chest CT scans were in February of 1975, using the ACTA scanner. Images were 7.5 mm thick and used a raster of 160 x 160. Images were displayed on a 19-in color monitor, photographed directly with a Polaroid camera, and the snapshots were archived. Later, in 1976, others showed similar images obtained with EMI equipment (Figs. 1C, 5 and 6). Predictions about the value of chest CT scans were mixed, ‘‘Considering that the scanning cycle takes about five minutes, one would expect that scanning of body parts containing moving organs, such as the chest and abdomen, would present a problem. However, we have been favorably impressed in particular by the quality of chest films obtained. Clear visualization of the lungs, heart, and mediastinal formations was possible.’’ Others expressed some doubt, ‘‘In the thorax, CT scans rarely surpass the diagnostic accuracy of conventional radiologic studies.’’

‘‘NOW MY LIFE HAS CHANGED IN OH SO MANY WAYS’’



Stacks Image 105


Fig. 3. Patent diagram of Ledley’s 1974 ACTA scanner provided for ‘do it yourselfers.’ (U.S. patent #3,922,552; Nov 25, 1975.)

As this new expensive equipment spread, governments and insurers attempted to limit purchases. Certificates of Need were required in many states before a CT scanner could be purchased. However, the undeniable value of CT scanning—for both diagnosis and patient management—eventually overwhelmed attempts to control dissemination.1 In the United States, in 2000, it is estimated that over 62 million CT scans were performed annually. Today, approximately 30% of CT scans are of the thorax. Initially, most scanning was done for lung and mediastinal disease. With the advent of helical scanning, vascular and cardiac imaging has become a major focus. With each technical advance, new CT scan applications arose.

Stacks Image 123


Fig. 4. (A, B) Translate-rotate detector versus fan beam detector. Original legends: ‘(A) Moving fan beam with rotation system. Fan beam produced from X-ray tube falls on X-ray detector and scans backward and forward, linearly, across the patient. At each scanning stroke, angle of scanning traverse changed by an amount equal to the angle of the fan beam. In this rotational method, all possible angles of scan across entire body will have been recorded after a 180 degree rotation. (B) Rotating fan beam system. X-ray tube produces fan beam as wide as large patient to be scanned. This was on wide array of 300 detectors. To obtain picture, assembly rotated around patient 360 degrees.’ (From Hounsfield GN. Picture quality of computed tomography. AJR Am J Roentgenol 1976;127:3–6; with permission.)


Stacks Image 121


Fig. 5. Original legend: ‘CT scan of a 79-year-old woman who had subtotal thyroidectomy, three years earlier, for papillary adenocarcinoma. Destructive lesion had recently developed in the right third rib and in the right side of the pelvis. (A) Tomographic section through the chest revealing large destructive lesion in rib on right. Large potion of the mass projects into the chest, and tumor extends considerably into chest wall. (B) Lung parenchyma revealed after adjusting window height and width. Note tiny nodule in apex of right lung anteriorly, which was not appreciated on chest film.’ (EMI Body Scanner) (From Sheedy PF 2nd, Stephens DH, Hattery RR, et al. Computed tomography of the body: initial clinical trial with the EMI prototype. AJR Am J Roentgenol 1976; 127:23–51; with permission.)


Lung

Initial scanning concentrated on parenchymal consolidation and tumor, as the peripheral anatomy of the lung was not well demonstrated on 10 mm slices with long breath holds (Figs. 1C, 5 and 7). With improved equipment, thinner and thinner scans became available. The first English language report of high-resolution CT (HRCT) scanning (1 mm slice every 10 mm, high- resolution algorithm) was in 1985. HRCT became the hot new area in the late 1980s and 1990s. Multi-slice CT scanning now permits one to obtain sub-millimeter images using isotropic voxels and to perform high-quality multi-planar reconstructions. Now, lung detail and distribution of disease can be assessed at the sub-millimeter level in any plane. HRCT scanning has led to new under- standing, classification, and reclassification of various forms of interstitial lung diseases and bronchial diseases. HRCT scanning has dramatically changed the way radiologists, pathologists, and clinicians diagnose and understand interstitial lung disease. As with most advances, as some areas become better elucidated, new questions arise.

Stacks Image 179


Fig. 6. Original legend. ‘(A) Currently asymptomatic 57-year-old with known multiple myeloma. A chest radio- graph demonstrating 4 cm mass (arrow) in right posterior mediastinum. (B) CT scan showing the posterior medi- astinal mass arises from contiguously destroyed posterior aspect of the right fourth rib (arrow), a manifestation of her known multiple myeloma.’ (EMI Body Scanner) (From Stanley RJ, Sagel SS, Levitt RG, et al. Computed tomography of the body: early trends in application and accuracy of the method. AJR Am J Roentgenol 1976; 127:53–67; with permission.)


Stacks Image 181


Fig. 7. CT scan on early fan beam scanner. Scan rota- tion time was 9.5 seconds and the slice thickness was 10 mm. (GE 5000 Scanner.)


Mediastinum

CT scanning was immediately embraced as the method for evaluating the mediastinum. Conventional radiographs and conventional tomography often failed to detect or adequately characterize mediastinal pathology because of lack of contrast between the normal structures and the pathologic structures (Figs. 5 and 6). Because the mediastinum is less prone to respiratory motion, and intra- venous contrast (administered by drip) provided sharp contrast boundaries, CT scanning rapidly became the modality of choice. It was soon touted as both sensitive and specific for staging lung cancer metastasis. With time, it was realized that CT scanning was an improvement, but fallible, for cancer staging. The recent fusion of CT and positron emission tomography scanning has over- come many of those problems.

Fast-drip intravenous contrast infusion was a valuable asset in distinguishing the major vessels from other structures, but little thought was given to using CT scans for evaluating the vessels beyond the superior vena, cava, and aorta. As axial imaging became faster, more detailed vascular evaluation became possible.Early studies of the aorta for trauma, dissection, or aneurysm were limited by motion and had limited acceptance in the radiologic and the surgical community. Aided by power injection of contrast, single- detector helical and then multi-detector helical scanning, CT scanning is now the procedure of choice in the majority of patients with suspected aortic disease.

The first CT scan report on pulmonary embolism, in 1977, looked at parenchymal changes but did not even mention the visualization of intra- vascular clot. The first mention of prospective evaluation for pulmonary embolus involved three cases using ‘‘rapid sequence of up to twelve 2.4- second scans, with a one-second delay between scans’’ (Fig. 8). Although large pulmonary emboli could be visualized with axial CT scanning it was not until the advent of helical CT scan that large and midsized pulmonary vessels could be imaged routinely. Single-detector helical scanning al- lowed one to scan only 12 cm of the chest in a 24-second breath hold at 5 mm intervals. Multi- detector scanners shortened the breath hold to a few seconds, provided sub-second rotation time and sub-millimeter resolution. Helical CT scanning rapidly replaced perfusion scanning and angiography as the clinical procedure of choice and is now the gold standard.

The same improvements, with 16-detector scanning and above, have made cardiac imaging and coronary artery imaging possible. Elegant 2- and 3-dimensional (D) reconstructions add a new dimension. CT scanning’s role is yet to be defined, as there are many competing modalities.

Stacks Image 167


Fig. 8. First published prospectively diagnosed pulmo- nary embolism. 1977 Axial image using a fan beam/ rotating detector with 9.5 second gantry rotation. (From Godwin JD, Webb WR, Gamsu G, et al. Computed tomography of pulmonary embolism. AJR Am J Roent- genol 1980;135:691–5; with permission.)


Lung Cancer Screening

CT scan screening for lung cancer is another recent addition to the diagnostic armamentarium. High-end scanners provide relatively low-dose images, capable of detecting even the smallest cancers. Unfortunately, other clinically irrelevant nodules are seen with great frequency. The eventual role of lung cancer screening is still being debated.

Other Developments

Many of the high-end applications discussed above are possible because of high quality 2D and 3D reconstruction. The development of the isotropic voxel has provided high-quality volumetric scanning and reconstruction in any two-dimensional plane or as three-dimensional images. Lung applications are numerous, including bronchiectasis and interstitial lung disease. Mediastinal and vascular applications include the trachea, the aorta, the great vessels, the pulmonary vessels, and cardiac applications—especially when gating is applied—include cardiac and coronary evaluation.

Early CT scans of the chest were 10 mm each. A chest CT scan consisted of 15 to 20 mediastinal and 15 to 20 lung images displayed on film. Now, with multi-detector CT scans and routine reconstructions, such as coronal, sagittal, or maximum intensity projection pressure images, the chest CT scan is often well over 1,000 images. Picture archiving and communications systems have made it practical to view the staggering amount of data that each case presents. Computer-assisted diagnosis (CAD) offers a possibility of reducing the information overload confronting the radiologist on a daily basis. Sophisticated nodule detention and nodule quantization software promises to make these two tasks less burdensome. In addition, CAD programs, now in testing, can also detect pulmonary emboli to the sub-segmental level and quantify emphysema.

‘‘WILL YOU STILL NEED ME . . . WHEN I’M 64?’’


As this article is written, the 64-slice multi-detector scanner is the dominant CT scanner. Although it is a current workhorse, the industry is rapidly moving forward.

Phillips now has a functional 256-slice multi-detector scanner that can provide images of the entire thorax within seconds and of the heart, in less than 2 beats. Toshiba (320 slice) can image the entire heart in one rotation. These newer scanners show the most promise for cardiac and coronary imaging and other small organ imaging. Technical limitations currently limit more general use.

Dual-energy scanning also has multiple potential applications in the lung. With dual energy, calcification can be more easily assessed and perfusion scanning of the lung is possible. This has potential applications for evaluation of pulmonary emboli and other clinical scenarios where perfusion information is helpful. Unlike nuclear studies, it provides both anatomic and functional information in the same scan. Myocardial infarct imaging, and perhaps coronary calcium removal, may be in the future. Dynamic respiratory imaging for airflow obstruction, tracheobronchial mechanics, and diaphragm motion are also possible with faster scanners.

Summary

Chest CT scanning has come a long way since 1975. Anatomic images are now superb and func- tional imaging is in its early stages.

Acknowledgements

Thanks to Dr Stanley Fox, PhD, for his historical insights, and to Mrs Sylvia Bartz, for her help in preparing this manuscript.