General introduction

The good radiologist has been trained to seek those features that normally lie outside of the awareness of regular folk in the real world. Originally the human visual system was evolved to enable us to make very fast assessments and then to get out of the way of danger that was seeking us. The processing involves compromises that are not always appropriate to Radiological assessment.

Radiographic interpretation is a systematic search for patterns of disease, that is independent of the image display or noise in the image.

Most approaches to the teaching of radiographic interpretation involve a combination of systematic analysis and looking for remembered patterns. The good Radiologist has to be consistent and training involves at least one imposed scheme. It may be boring to go through a list each time but initial ennui saves money in later indemnity charges.

For each radiographic projection, the beginner is advised to have a list of features that they will deliberately observe. Repetition will convert this to an automatic response, rather similar to the concert pianist who does not have to be deliberately aware of fingering, but can concentrate on musical interpretation.

The George Simon technique was to have his juniors go through a pile of normals, while making note of a number of features. Remember that the process is a means to an end and not an end in itself. Basic pulmonary structure and function is discussed elsewhere. Knowledge of lobar anatomy can be gained by study of the different patterns of atelectasis.

Salience

Radiologists soon learn that their visual neurophysiology is not pure and unbiased. All trainers in any field that involves observation are aware that if something is unnoticed after a certain time, the human will be unable to see even a feature of high salience. 'Salience' is a way of saying that the observed structure comes into the usual range of contrast and spatial frequency or sharpness; in other words, how easy it is to see the thing.

An example of an object of high salience would be greater than 10 minutes of arc at the retina and would be a black on white image. The low salience object would be of low contrast difference with its surroundings, unsharp and with a confusing background pattern. Eye saccidic movement studies show that beginners waste a lot of visual effort on structures like the heart and great vessels that have clearer margins.

The visual system extracts features from an image and the process can be fatigued by a prolonged stimulus. Some illusions are based on this property. The argument is not limited to low order neurological mechanisms. Analysis can be extended to extracting the signal of pure mental effort from the noise of other worries and personal insecurities. There has to be a reason why accidents are more likely to happen to a car driver within 10 minutes of making or receiving a mobile telephone call.

If you are demonstrating a film to someone else, never point Physiologists note that you would look more at the end of your finger than the film and it upsets the gestalt or unconsious assessment of the film features for both yourelf and the person to whom you are demonstrating. The parietal lobe processing can be left to the 3D task of anatomy assessment. Tasks in real or mentally-constructed three dimensional space interfere with each other, if done simultaneously. A verbal task will not interfere. This explains why Radiologists prefer to dictate, rather than type their reports.

Another point that is ( unfortunately ) relevant to exams and presentations is that a learned task is managed in a different part of the brain from the learning area. If from insecurity or under pressure you do a previously well trained task carefully and deliberately at a conscious level, you are using the 'learning' area and may degrade your performance, ask any golfer.

In brief, there seems to be a limited amount of observation, activity and interest available to the observer regarding a particular picture. Spend it wisely!

It is inappropriate to impose a particular scheme upon those who will interpret a chest X-ray and the following is only intended as a guide. After all, my neurophysiology is probably different from yours.

One useful approach to the film mimics the approach of a clinician, who will observe the habitus, movements and respiration of the patient. An important part of neurological assessment, for example, is to watch the patient as they walk into the consulting room and get on the couch. The Radiological equivalent would be to note any rotation of the patient to the film and then to assess any asymmetry, particularly low contrast or low spatial frequency assymetry. This will sometimes lead you to the part of the film which requires detailed observation first.

The analytical method will demand that you try not to seek pathological explanations before making all your observations. A small chest volume might be due to weak muscles rather than stiff lungs and the thymoma may not be obvious. As you get more experienced, you will be able to separate those images that are completely normal or unequivocally typical of some common disorders. Nevertheless, some deliberate review of film features will be necessary to inform your discussion of any differential diagnosis.

Once you have made the observations, go for the sign with the shortest list of differential diagnosis and choose the diagnosis compatible with the other signs and the underlying probabilities in that patient group. For example; apart from more consistent exposures and greyscale, one reason why digital radiography has been accepted in the Intensive Care Unit is that the situations there are often predictable consequences of assisted respiration and the sub-set of procedures or disease processes that resulted in a transfer to ITU. The close clinical supervision means that the lower resolution image only has to help the decision between a more limited number of possible conditions.

Remember that occasionally, the probability of a combination of two common conditions might be greater than the probability of one rare condition. The reason for being systematic and delaying the leap to the diagnosis is that it requires much more information to shift an erroneous opinion than for the correct answer to reveal itself to the unbiased mind.

Lung volumes

Lung volumes will have a huge effect on the appearance of the chest X-ray, especially in babies where the greater relative movement of the diaphragm will distort the trachea in expiration.

The rise of the carina on expiration will influence what is the safe limit for insertion of an endotracheal tube, which should be above the carina at all times.

Symmetry

Deliberately look for features of low contrast and low spatial frequency This is another way of saying 'look at the film from as if from a long distance away'. The old practice of using a minifying mirror or lens to examine a radiograph underlies the usefulness of this argument. This feature might be a way of solving gradual image delivery in the all-digital hospital, but involves argument about Gestalt and manipulation of observer performance by the display medium, a very tricky field that is best left out of this discussion.

Any low contrast assymetry will sometimes lead you to the part of the film which requires detailed observation first.

		This image is used to show the radiological importance of low contrast and low spatial frequency. The eyes of the beginner are always drawn to the splenic and coeliac node calcifications. The calcific density of the shrunken left kidney is then seen. Some will notice the absent right breast and the radiation change in the right lung, but they never see the wide zone of transition of the destruction of the outer margin of the right ilial metastasis, made even less visible by the resolution of this image.
		The patient is female and the slight asymmetry is due to the simple mastectomy on the left and the radical resection on the right side. The abnormality is low spatial frequency and low contrast. It is difficult to see because the edge of the remaining pectoralis major is hidden by proximity to the right clavicle.

Camouflage, false negatives

This brings me to the discussion of camouflage. The zebra looks bizarre against a plain background, but is coloured to reproduce the predominant contrasts and spatial frequencies in tall sunlit vegitation. Some radiographic features are obscured in the same way.

		Beginners might detect the mediastinal lymphadenopathy by the replacement of the right tracheal stripe by a right para-tracheal mass, but the largest pathology imitates the size and shape of the heart, making the huge lower mediastinal lymph nodes less obvious.
		This post-op chest provides another example. The right lower zone band is not a rib. It is formed by a collapsed middle lobe, bounded by sub-diaphragmatic air below and the horizontal fissure above. Systematic examination of the hilar vessels reveals that the basal artery has disappeared. It has disappeared into a triangular density at the right cardiophrenic angle, a collapsed right lower lobe. This is the reason for seeking the hilar point. Provided the student is consistent, the usefulness of the concept lies in aiding the detection of odd collapses and hilar distortions.
		I'll give a further example to emphasise the point that camouflage can be responsible for lots of errors in film reading. The radiation change over the left hilum is easy enough to notice, but the presence of metastasis in the left 4th and 5th ribs is obscured, because the densities superficially imitate the appearance of normal scapula and ribs.

Illusions and false positives

		click on the image to see the illusion. A skin fold projected over the lungs can imitate retinal edge enhancement and give a false impression of a change in greyscale between two fields.
		This is an example of the Cornsweet illusion that sometimes confuses Radiologists.
		There is a density to the right of the clavicle. This lies above a visible azygous vein. The normal azygous vein lies at the right tracheobronchial angle, running forward from the para-spinal hemiazygous system to the superior vena cava near the front of the mediastinum.
		A normal variant is for the developing azygous vein to nip in a portion of the developing lung bud to make an azygous lobe. The four layers of in-drawn pleura are visible as a curved line.
		The solidity of the density is abolished by concealing its linear edge with a dark line. The appearance of solidity is partly conferred by the curvature. Cartoon characters in a newspaper depend on this response and ultrasound assessment appears complex to beginners for the same reason.

Experienced radiologists will remember many instances where seemingly complete lesions and ring shadows turn out to be combinations of overlying structures. Although less critical than the bright light and a magnifying glass in the plain-film days, some digital manipulation of contrast and image-size remains appropriate to make some pathologies visible. Feature extraction software may help to eliminate false negatives, but leaves the Radiologist with the problem of evaluating the false positives.

Vessels

		The deliberate search for the 'hilar point' helps to avoid the missing of lobar collapses. The disappearance of an expected vessel into a triangular density is similar to the invisibility of the ribs of a closed fan held by an angry 18th century lady. The vessel count in the ( remaining ) aerated lung is reduced. The upper mediastinal lymphadenopathy suggests the underlying pathology of bronchial carcinoma.
		Conversely, if you can see the basal artery, there isn't lower lobe collapse or consolidation. The lower density is due to an effusion in the greater fissure.
		If you were designing a cybernetic radiology system, which could only look at one part of the lungs, then the best place to choose for the pattern analysis would be the right lower zone. There, the veins cross the arteries at the steepest angle on their way to the left atrium.
		The practice of following the vessels out to the periphery means that you will give yourself a second chance not miss this pneumothorax.
	One important point about any vessel whose contents are under pressure is that 2 or 3 dimensional elastic structures when subject to a force will stretch in two or three dimensions.
		This applies particularly to the aorta, which is anchored securely by the intercostal vessels and starts or finishes at fairly reproducible locations. When the aorta gets wider, it elongates and becomes tortuous. Now you know what the radiologists mean by "unfolded". You may also see extreme tortuosity in pulmonary vessels that lead to an arterio-venous malformation in the lungs.
		These portal veins in cirrhosis are indicated by curved lines of calcification. Tortuous, dilated ureters are another example. Now, you may understand the downside to dietary fibre. Those capacious African colons are also longer, which may explain the greater risk of colonic volvulus in Africans.

Blood flow

Changes in pulmonary perfusion and non- uniformity of the lungs from disease can result in confusing appearances, particularly with emphysema.

		The only near normal region of the lungs is in the right lower zone in portions of right middle and lower lobes. These 'marker vessels are in the region that will develop pulmonary oedema first.
		The abnormal lung is 'protected' by its effect on local pulmonary perfusion.
		Interstitial pulmonary oedema shows the basic pulmonary units at the lung edge. Kerley 'B' lines. Not all lines are pulmonary oedema.

Extrapulmonary pathology

		Extrapulmonary pathology is not always as easy to see as this case of diaphyseal aclasis, multiple exostoses.
	The normal chest X-ray is taken with rays from a point source behind the patient with the film in-front. The central beam is on the 6th thoracic vertebra. Given that the centre of radius of the thoracic spine is somewhere in front of the patient, the geometrically gifted can infer that the end-plate margins of the vertebral bodies are not visible except the lowermost and that the upper thoracic laminae are best shown in the upper part of the film.

The Lawyer zone

The casual observer can often miss the absence of parts. This is nowhere more true than at the lung apices and upper mediastinum called the lawyer zone by Ed Gurney. The overlying structures give a confusing sense of normality. See also Camouflage.

		The destructive process in the posterior part of the right 3rd rib is obscured by the densities of the right clavicle and right 1st rib. An even more subtle destruction in the superior border of the anterior end of the left 2nd rib exactly corresponds to an expected lucency between the posterior margins of left 4th and 5th ribs.
		In this patient with lymphoma infiltration, the right apical density has an curved inferior margin and ends too high for the innominate artery in a young patient with a normal aortic arch.