The EMI scanner (1975)

Heat. Heat. [Music] [Music] [Music] This talk is to do with the new method of investigating the brain. But first of all, I think I ought to put the method into perspective by telling you what may happen to a patient entering hospital for investigation of disease of the central nervous system. And for the sake of simplicity, we're going to suppose that our patient has a tumor somewhere in the right hemisphere, probably in the right parietal lobe. Now, before we can institute treatment, we must know exactly where that tumor is and if possible, what it is, whether it is a primary tumor or whether it is a secondary malignant tumor from a site elsewhere in the body or whether it is benign. Now in addition we have a whole battery of investigations which we can use and we must choose these with due care for our patient. Some of these investigations are simple and non-invasive whilst others involve the use of complex invasive techniques and there is always the outside chance of a serious complication occurring. The simple techniques include straight skull radiography, radionuclide scanning and ultrasound. Whereas the more complex invasive techniques may take the form of either cereble angography or cereble pneumography. If I were a patient, I think I'd be scared to death death at the mere thought of having any of the last two mentioned investigations performed on me. Now it is an unfortunate fact that each of these investigations allows us only a very restricted look at the structures which comprise the whole cranium. Skull radioraphs show us the bony structures in various projections but give us no indication of the soft tissue structures contained within. The conventional system of X-ray source object and sensitive film ensures that all the information relating to differential absorption along the path of the X-ray beam will be superimposed on a particular place on the film. The dense bony confines of the skull therefore have a blanketing effect and we get little or no indication of the intraraanial anatomy. We may be restricted to seeing only the displaced pineal body when it is calcified. Radioisotope scanning shows us where the administered isotope has collected in abnormal concentration but again it gives little indication of the anatomy existing within the cranium. Ultrasound shows by echolocation the position and form of interfaces within the brain. We may thus be able to say whether the midline or adjacent structures are displaced or deformed or whether the interfaces arising from abnormal conditions within a mass such as a hematoma or a tumor. We again do not get a direct look at the soft tissue structures. Arterography enables us to look at the vascular structure of the brain and from displacements, change in contour, pathological changes in fine vessel structure or disease of the vessels themselves, we may be able to deduce the sight and probable nature of the lesion. Cereble pneumography enables us to see those structures which are normally filled with fluid that is the ventricular system the subaractid basil systems the cereal salsi and again from changes in position size and contour we may be able to deduce what has happened. It requires no great stretch of the imagination to see that what we have required for a long time now is a system which would enable us to look at the structure of the brain without causing the patient fear or discomfort. This is the EMI scan of the patient whose skull X-rays and technesium brain scan we have been looking at. It is now nearly 80 years since Runan discovered X-rays and it is true to say that in that time the system of X-ray source object and sens sensitive film has changed but little. There is however a way in which we can change the system. Let us imagine that we are going to look at a cross-section of some part of the body in our case the cranium. This will be a finite thickness and we are going to irradiate it with a columnated beam of X-rays. Over the thickness of the slice, we can imagine that the beam will be nearly parallel and therefore of nearly constant dimensions. By using a system of crystal detectors, we can measure the number of photons entering at a particular point and the number of photons emerging at the corresponding exactly opposite point on the circumference. By mounting the X-ray tube and crystal detectors in a common frame, the columnated X-ray beam can be made to scan along the plane of the slice. During this movement, the irradiation of the slice is regarded as a continuous sequence of beam paths or X-ray transmission profiles. Along the path of the beam, X-ray photons will undergo exponential absorption. And if we regard the slice as being made up of a number of parallel adjoining beam profiles, we could, if we were able to calculate the absorption coefficients of equal individual segments of each beam path, effect a reconstruction of the internal structure. A total of 240 equally spaced readings are made in each linear scanning motion of the X-ray tube. This is then followed by a onederee rotation of the whole gantry around the circumference of the slice and the scanning motion is repeated and so on until a rotation of 180 degrees has been completed. In 1968, Mr. GN Houndsfield, a research engineer at EMI, effected a solution to this problem. But before I go on to describe the apparatus which we call the EMI scanner, I would like to mention the names of three people who have looked at this problem in different ways but always with the same objective in mind. The first is a Los Angeles neurologist WH Oldenorf who in 1961 devised an experiment which led him to believe that he might be able to calculate absorption values along the path of a beam of X-rays passing through the head. He did in fact reach out towards the solution but didn't quite make it. The second is a physicist am Cormarmac who in 1963 was experiment was experimenting with X-ray transmission profiles and found that he was able by back projection to accurately calculate the absorption values of the pure aluminium and oakwood portions of his model. The third person was dee cool who in 1968 after a number of years of experimenting with projectional methods finally solved the problem and created an effective uh transverse sectional radionuclide brain scan. Now let us look at the EMI scanner and see if we can explain the method employed with this apparatus. The cranium is regarded as a series of contiguous slices, each made in the transverse axial plane. Each slice is scanned in turn using a highly columnated X-ray beam and an equally columnated system of crystal detectors. Projections are obtained by measuring the attenuation of X-ray photons across the slice from a large number of sequential exactly corresponding points around the circumference. A computer is used to assemble this information and arrange it in a form suitable for use in a series of mathematical operations. This series of mathematical operations or algorithm is then used to effect the reconstruction of the anatomy of the slice. Reconstruction is made by regarding each slice as a matrix made up of 160x 160 cells. Each cell represents a block of tissue 1.5x 1.5x3 mm. The last dimension 13 cub mm is the thickness of the slice. Using back projection methods, an absorption value is calculated and related to its correct cell in the matrix along the corresponding beam path or projection. The computer registers 43,000 readings of X-ray photon transmission. And it is from these readings that a sufficient number of simultaneous equations can be constructed to enable the computer to calculate the necessary 25,600 absorption values for the matrix of 160x 160 cells. The absorption values are used as picture points in a cathode rate tube intensity display based on the relative densities of different tissue structures. A convenient scale of absorption values has been devised so that there is a range of 1,000 units between the least dense normally encountered substance which is the air in the paranasal sinuses and mastoid air cells and the most dense substance which is the compact bone of the skull. Air has an absorption value of approximately minus 500 and compact bone an absorption value slightly in excess of plus 500. The soft tissues of the brain have a fairly low density and vary between plus one and plus 22 or approximately 4% of the positive range. The most dense tissue that is bone is depicted as peak white while the least dense tissues air and cerebrros spinal fluid are depicted as black. Soft tissues of the brain are depicted in shades of gray shading from dark to white with increasing density. By operating the appropriate controls, the operator can select a display which gives maximum detail for the tissues under examination. The operation of the scanner differs from that of conventional X-ray equipment, but is essentially simple. The X-ray tube, which is continuously operating, requires a fairly lengthy warming up procedure, similar to that followed with X-ray therapy tubes. The patient's head must be correctly located in the machine, and for purposes of ease of calculation and accuracy, must be placed in the center of an exactly square perspect box, which is filled with water. This is achieved by forming the front face of the box with a rubber diaphragm which can be drawn in to form a cap in the center of the box. When the patient's head is located in the box, the cap can then be allowed to shrink onto the patient's scalp, thus eliminating any air gap. A simple measuring device is then used to locate the correct level of the scan. And once the computer, X-ray control console and scanning gantry have been correctly integrated, scanning sequence can begin. It takes about 5 minutes to complete, during which time the patient must remain perfectly still. Pictures of two contiguous slices are available approximately 20 seconds after the completion of the scan. A complete examination from base to vertex usually requires four complete scans that is eight pictures. Now let us consider the results and their interpretation. The pictorial display is examined in much the same way as a conventional radioraph. Structures are identified and their shape, size and position defined. Changes in absorption are then looked for. The perspective differs from that to which we have become accustomed in conventional radioraphs and we may have to cast our minds back to our medical student days when we were accustomed to looking at transverse sections of the brain. However, the importance of the display is that living anatomy may now be shown quite simply and with no particular discomfort to the patient. In the posterior fossa we can see the fourth ventricle the brain stem the pontine system and the lateral extensions into the cerebellop pontine angles. Coming to the next slice up we can see the quadrogeminal system. We can see the thin slit of the third ventricle and the sylvvian fissures situated more laterally in the temporal fossy. Up one more slice we see the anterior horns of the lateral ventricles, the bodies of the lateral ventricles, the third ventricle, the pineal body, calcified coroid plexuses, and the occipital horns and triones of the lateral ventricles. Situated more laterally, we can again see the slits of the sylvian fissures. And between these and the lateral ventricles, we can see the complex structure of the corporus triata. Still coming up we see the top halves of the lateral ventricles again with the coroid plexuses and the third ventricle. And coming still higher we see the extreme parts of the roof of the lateral ventricles with the corpus corpus collars in between them. In the topmost cut we run we've come up above the major form structures and we now see the homogeneous density of the white matter. And in these scans unless there is an abnormality present present we can't really distinguish cortex from white matter. But if there is edema or if the patient has cereble atrophy causing widening of the cereal salsi we can distinguish these structures. Now let's consider the abnormal scan. In neurological practice, the common lesions which require identification are cereal neoplasms of all varieties, hematomas, infactions and infections. To this list must be added cereal edema which often accompanies these lesions and complicates the clinical picture as well as the findings presented by other methods of investigation. As I have mentioned before, structures are identified and their shape, size and position defined. Changes in tissue density are then looked for. Usually a space occupying lesion will produce a characteristic displacement or deformity of some part of the ventricular system. But for precise identification, the extent of tissue change needs to be defined. Tissue abnormalities may be conveniently divided into three large groups according to density alterations they exhibit in the scan pictures. Firstly, lesions with an average density higher than that of normal tissue. Secondly, lesions with an average density lower than that of the surrounding normal tissue. And thirdly, lesions with the same density as normal tissue. Now, high absorption values may arise out of deposition of calcium in different types of lesions. It may also arise out of low water content in a closely lit or fibrous structure or to a hemorrhage in which clotting has occurred. Low-grade astrocytoomas, oligodendroglyomas or appendimomas are examples of neoplasms which show up as white areas in computerized scans because of their calcium content and consequently high absorption values. The calcium aggregates may be in simoma bodies or calcospherite formations. Now this is a patient with an oligodendro and you can see the calcium aggregates as dense spots within the tumor. It has produced a deformity of the anterior horns of the lateral ventricles and must be clearly extending into the genu of the corpus colosum. The dark area is an area of edema. Now the next patient is a quite different problem. This is a patient with a benign tumor. This is a menioma and shows up because of the calcification in the simoma bodies. Behind the tumor you can see a roughly crescentic area of darker brain which is adeus. Other lesions containing calcium are equally well shown in hemorrhage. Once clotting has occurred, serum is progressively absorbed and the concentrated blood constituents will then have a much higher average absorption value than normal brain. Absorption values for hematoma may vary from plus 22 to plus 36 depending on the amount of serum absorbed. This large difference in absorption values between hematoma and normal brain enables the computerized scan to operate to great advantage in cranioal trauma or indeed any condition where a hemorrhage is suspected and the hematoma requires to be delineated. Now here we have a patient with an obvious extraural hematoma in the left posterior parietal area. You can see it characteristically inwardly convex edge and it shows the very high density of clotted blood. The lateral ventricle on the contrlateral side is seen to be displaced away from the midline. The next patient is a different problem. This was a man who was admitted to hospital deeply unconscious with a history of having been sick, fallen backwards and struck his head on the bathroom floor. The question then arose as to what sort of damage he had sustained to his brain. Did he have an extraural hematoma, subdural hematoma, an intracerable hematoma or was his brain merely confused? The EMI scan gives us the answer because here in the frontal loes we can see this patchy increase of density which looks like segments of cauliflower surrounded by darker areas which are obviously admitous brain. This is frontal lobe contusion. In fact, a contra injury. Primary intracerable hemorrhages are seen in a hitherto unobtainable perspective. And from a surgical point of view, the demonstration of the hematoma, its size, relationship to deep structures, and the point of nearest approach to the surface of the brain or the extent of surrounding edema may be very valuable. Now, here we have a patient with just such a hemorrhage. You can see the extent of it, what structures it has involved and where it comes closest to the surface. You can also see what the compression effect on the lateral ventricles has been. You can also see a thin strip of clotted blood lying subendomly along the wall, the lateral wall of the compressed right lateral ventricle. Here is another patient with a primary hemorrhage, but this time in the vermis of the cerebellum. Now this is an extremely dangerous type of hemorrhage as far as life is concerned and the sooner it's diagnosed and the compression relieved the better for the patient. Here you can see that the EMI scan has provided us with all the information we require. We can in fact uh perform an operation and remove the clot merely on this information. In other cases of hemorrhage, an ruptured aneurysm may be responsible. And if we have multiple aneurysms, it may be difficult or impossible using the means ordinarily at our disposable at our disposal to distinguish which aneurysm has in fact ruptured. This is where the EMI scan plays another important role. We can detect the small local hematoma. And this is just such a case. This was a patient with multiple aneurysms, including an anterior communicating artery aneurysm. Here is the hematoma lying immediately in front of the third ventricle, indicating that the anterior communicating artery aneurysm had in fact ruptured. Low absorption values occur in many different disease processes, including neoplasms, infections, infactions, or indeed any process where there is a breakdown of normal cell structure and or an increase in water content. For example, coagulative necrosis, microscopic and macroscopic cyst formation or gross deminating processes. Now in the scan, we can see a well- definfined area of low density. This is obviously a cyst filled with fluid and it's associated with a malignant tumor. Here we have another area of low density, this time less well- definfined. And in the center we can see some area of slightly increased density. This is a metastasis from a carcinoma of the broncus and the dark area surrounding it is in fact a demitus brain while this rather ragged irregular area in the middle is in fact the neoplastic tissue. In other scans we may see changes which we are unable to interpret in terms of precise pathology. Now in these two scans here we can see dark areas which may or may not indicate edema and we require to know a little bit more about the patient. Now in most cases we usually do for instance if this patient here had an empa of the lung which was drained and then later on became very ill developed a left hemiparesis we would suspect that she might have developed a met metastatic absis in the right cereal hemisphere and the EMI scan was in fact showing the edema. We can't actually see the absis cavity itself, but I'll come back to this particular patient later on. In the next scan, we see a very similar appearance this time affecting the left frontal lobe, but without much displacement of the ventricular system. And again, we require to know a little bit more about the patient. If this patient had had a sudden incident which resulted in a right-sided hemiplegia, we would say that he or she had sustained a stroke, possibly an embolism, uh possibly a thrombosis or even possibly a hemorrhage. Now, we know it isn't a hemorrhage because we can't see the density of the hematoma, but we see a large well-defined low density area and this indicates that the patient has in fact sustained an occlusion of a major vessel. And this is the infacted brain showing up as an area of low density. As one would expect, subdural hematomas vary considerably in consistency. Those hematas which have a thin watery consistency are seen as characteristic lowdensity crescent capping the cortex. Where the consistency is thick and tarry, the hematoma may be more dense than the underlying brain and shows as a dense crescent. Other subdural hematomas with the same average density as the underlying brain may give rise to difficulty in the diagnosis. The large displacement of the ventricular system and the unbroken apparently normal brain density should point to the correct diagnosis. Now here we have a typical lowdensity subdural hematoma. You can see the low density crescent capping the cortex on the left side. But there is more detail in this. If we look at the contrlateral hemisphere, we can see the cerebrospinal fluid lying along the fal cerebri on the right side. And we can also see some detail of the cereble salsi but none on the left side. This is an indication that the subacoid space on this side and the cereal are compressed. The third group of lesions exhibit average X-ray absorption values which are the same as the surrounding normal brain and are difficult to identify in ordinary scans. When the lesion is large enough to displace or to distort identifiable structures such as the ventricular system, the existence of a lesion is recognized, but its exact location may not be identifiable. Should the lesion be small and peripheral in location, it may not even produce any distortion or displacement and may therefore escape detection. However, since this the sensitivity and accuracy of the method is of a very high order, artificially raising the absorption values of abnormal tissue and enhancing the density difference of the boundaries of the affected tissues becomes a distinct possibility. The breakdown of the bloodb brain barrier in abnormal tissues allows small amounts of circulating substances containing heavy atoms to pass into the abnormal tissues and to be retained for relatively long periods of time. Sodium ioamate which is a available in most X-ray departments as a contrast medium can be injected in large amounts and appears to pass through the bloodb brain barrier mainly by a process of passive diffusion. The process is rapid and increased density may be seen in many cases in scans made 5 minutes after the introvenous injection of 60 ml of sodium or conre 420 which is its trade name. As the process is one of passive diffusion, the degree of the epacification is dependent on the concentration of circulating sodium. And as one would be one would expect, tumors with large vascular beds tend to show up more densely than those which are comparatively avascular or necrotic. In some disease processes, the lesion may not be seen in the ordinary scans and certainly not distinguishable from the surrounding edema which it has produced. In these cases, density enhancement of the abnormal tissue with sodium becomes very valuable. Now here we have a patient with a meningioma. This can be distinguished in the normal scan because of the edema behind it. But if that edade edema had not been there, one could imagine that this would not have been distinguish distinguishable from the normal brain. But give the patient 60 ml of Conra 420 by introvenous injection and then scan immediately after the injection. You can see the tumor very clearly. It has increased in density very very considerably and shows a very sharply demarcated boundary. This is an indication that it is probably benign. Here is another patient with a lesion which was difficult to identify in the normal scan. This was a patient who had a sudden vascular incident which left him with cortical blindness. Now the chances were that this patient had somehow or other oluded his posterior cerebral artery circulation. We gave him 60 milllers of Conra 420 by introvenous injection. And here you can see the greatly increased density of the infared tissue in the distribution of both posterior cereble arteries. Small tumors such as metastases may be identified and in others the extent of invasion or infiltration by tumor tissue may be seen. An abapsis cavity may be distinguished from the intense edema which it usually promotes in the surrounding tissues. Now here we have a patient with a typical malignant gloma. The density of the malignant tissue has been enhanced with conre and you can see how it comes backwards. It has greatly deformed the anterior horns are both lateral ventricles. It has invaded the genior of the corpus colosum and extended across the midline into the medial aspect of the left frontal lobe. Behind it you can see where there is soft adeus brain. Now casting our minds back to the patient who had had an empaire drained and who was suspected of having developed a metastatic absis in the right cereal hemisphere. You will remember that all we could see was a dematus brain. We could not really identify the absess. But having given the patient 60 ml of conre 420 or sodiumate we can now see it. It shows up as a dense ring. This is because the granulation tissue in the wall of the absis cavity has taken up the contrast medium. Now with attention to correct scanning technique and arranging scans to coincide with areas of interest, tumors in and around the pituitary fossa may be identified. Here is a patient with a large pituitary adenoma. In the first scan, we can just see the tumor protruding above the pituitary fossa and deforming the kaismatic system. It is not really very dense. But if we give the patient 60 ml of conra 420, the density is considerably enhanced. This is a value when the tumors are small or difficult to identify. And we could easily imagine a situation where we could see uh a small asymmetric tumor rising out of the pituitary fossa. When we come to the tectal plate area, this may also be examined in some detail. Tumors arising in this area will cause an obstruction. The lateral ventricles will enlarge and the third ventricle will become widened. The tumor itself may be partly calcified in which case uh it's a pinealoma or some sort of tumor which arises in the tectal plate area. Now the superpineal recess is usually obliterated and we can't see it. Here is just such a patient. Here is the tumor of slightly increased density occupying the posterior portion of the third ventricle obstructing the outflow of cerebrospinal fluid and causing ventricular dilotation. Now lastly with some slight modification of technique the contents of the orbit may be examined. The optic nerve is surrounded in the muscle cone by fatty tissue and as this has negative absorption values on the scale used. Any alteration of density due to abnormal tissue may be easily identified. Here we have a patient with a tumor which is causing proptosis of the right eye. If you cast your eyes on the left side, you can see the globe of the eye. You can see the optic nerve. You can see the thin strap of the lateral rectus muscle and the unbroken low density of the fatty tissue within the muscle cone. On the right side, the situation is altogether different. The optic nerve is straightened and displaced medially. And here in the outer lateral quadrant of the eye, you can see a large dense tumor. This is a cavernous hemangi. Now in conclusion, since the EMI scanner is able to provide a demonstration of the intraraanial anatomy quite simply without danger to the patient and with no particular discomfort, it clearly has an important part to play in the investigation of conditions which are not clearly defined and which in the ordinary course of events would require at least one contrast radiological procedure. The apparatus which we have we have been describing is basically the prototype design and is therefore very much a first generation scanner. Improvements will no doubt be made in the directions of speed, accuracy, sensitivity, resolution, and the extension of computer aided operations such as light averaging uh sorry, light pencil averaging of suspect areas, data storage, classification, and comparison with previous experience. Application of the method to the thorax and abdomen has from the very beginning excited great attention. But there are formidable obstacles in the way. X-ray dose, sensitivity, matrix size, density, range and voluntary and involuntary movements are just some of the obstacles which must be overcome. This makes the design of an apparatus much more complex. This however should not now be beyond the ingenuity of modern technology. Thank you. [Music] [Music] [Music]