It is well documented in literature that all radiologists make mistakes when interpreting imaging studies [1] . Such mistakes may arise from radiographic artefacts which are structures not naturally present in living tissue but of which an authentic image appears on a radiograph [2] . They mask true abnormalities, create pseudolesions [3] which are distracting and compromise accurate diagnoses [4] .

Radiographic films are processed with the automatic or manual processors. It is documented that radiographic artefacts occur more commonly in the Darkroom [2] and with manual than the automatic processor [5] . Despite the introduction of digital radiography which has supplanted film-screen systems, artefacts have still not been eliminated but only reduced [6] . In order to avoid misinterpretation, recognizing artefacts and understanding their physico-technical background are of great importance in imaging [7] .

A review of literature reveals artefactual classification based on causative agents although artefacts can also be categorized by the mechanism of interference with image acquisition, processing, or display [8] . In a work done using mammography, a closely related modality to X-Ray, Van Ongeval et al. [7] classified artefacts as patient-related, technologist-related, machine-related, processing-related and viewing conditions-related. In another similar work four categories were discussed: machine, patient, technologist and processor [3] . Minus and plus density as a means of describing artefacts has also been used by Akinola, et al. [2] .

The weakness of the earlier classifications above lies in their inability to indicate specifically how and when artefacts are introduced into radiographs. The classification we propose is a condensation of stages of the radiographic process as well as specific artefactual activities and agents.

This was a prospective study carried out in the Radiology Department of a University Teaching Hospital in Nigeria between June 2011 and June 2013. Formula was used to establish a minimum sample size of 400 radiographs with artefacts. Using purposive sampling method, radiographs produced between June 2011 and June 2013 which were archived and viewed by the researchers using a 100 × 50 cm giant viewing box. Artefactual ones were subsequently separated from those free of artefacts. The artefactual radiographs were further scrutinized to determine the specific artefacts on them. This scrutiny was achieved by observation of each stage of the radiographic cycle as well as darkroom simulations of artefacts when ambiguity was high. The number of trends were subsequently differentiated. Simple statistical tools were used to calculate central tendencies and frequency.

Table 1 shows the characteristics of the isolated artefacts. A description of each artefactual trend, their appearance as well as specific causes are given. The frequency of occurrence is also given. Multiple-dispersed dots (35%) caused by dirty intensifying screens are the most common while grid lines (0.8%) caused by immobile or wrong surface of stationary grid has the least frequency. The proposed classification is shown in Table 2.

Our work suggests a classification based on six stages of the radiographic process. The appearance of artefacts in each stage is also given to aid specific decoding of individual artefacts.

Our study which is a combination of earlier works is strong on mnemonics. We suggest that artefacts be re-classified into packaging, procedure, patient, pre-processor, processor and post-processor.

Packaging in our context involves artefacts induced in the film emulsion by suppliers as well as end-users who store the films before exposure. We see procedure as every manipulation of the patient, machine and accessories done by the Radiographer in carrying out the radiographic examination. We noted pre-processor as the time interval between the exposure of the film to X-Ray and feeding into the processor. Post-processor we acknowledge as everything that occurs as soon as the radiograph comes out processed from the processor.

We observed that artefacts appeared as either dark or greyscale. Artefacts from patients and post-processor were found to hover between white and light grey, described as greyscale while those from procedure and processor went from one extreme of dark to the other extreme of white. Packaging and pre-processor artefacts were however, always found to be dark. This is probably due to the activation of silver halide by pressure which added to the overall activation by radiation, thereby creating a higher density. Radiopacities and dirty intensifying screens produced whitish-grey artefacts. It is strongly suggestive of significantly attenuated radiation leading to diminished silver halide activation.

We also observed that pressure and visible light on films before and after exposure to radiation caused dark artefacts. But pressure had no effect after processing. Only abrasion with a rough surface caused scratches and these appeared greyish. We recommend that our proposed classification be adopted as it addresses the genesis of artefacts in the film as well as the concomitant appearances.