The Importance of Patient Positioning in Radiography When Utilising Automatic Exposure Control

ABSTRACT

Introduction

Automatic exposure control (AEC) is a device used in radiographic practice which terminates a radiographic acquisition when a certain exposure threshold is reached. The AEC is commonly utilised for in‐bucky imaging, such as for the chest, abdomen, pelvis and spine. There are many benefits to utilising the AEC where awareness and knowledge of the technology are thoroughly understood. Poor technique when utilising the AEC, however, will alter dose, and by extension, image quality. This study aimed to assess how changes in patient positioning when using AEC for chest, abdomen, lumbar spine and pelvic x‐ray examinations affect patient dose and image quality?

Methods

An anthropomorphic phantom was marked every 1 cm from the ideal centring point in each direction; superior, inferior, left lateral and right lateral, for up to 5 cm of off‐centring. Exposures were taken at each position and the tube current, air kerma and signal‐to‐noise ratios recorded for each exposure.

Results

Air‐kerma changes ranged from a maximum reduction of 6.06%, 74.72%, 76.07%, and 35.66% for the chest, abdomen, pelvis and lumbar spine, respectively, to a maximum increased air kerma of 50.09%, 8.00% and 10.32% for the chest, pelvis and lumbar spine. The abdomen did not demonstrate any increased air‐kerma results. Statistically significant changes (p < 0.05) to dose were observed in 54 of the 80 (67.5%) separate recordings. Changes to signal‐to‐noise ratio varied from a maximum negative change of 49.14% to a maximum positive change of 25.10%.

Conclusion

Statistically significant differences in dose were commonly discovered for mis‐positioning beyond 1–2 cm, which will likely occur within the clinical environment. Such variation in dose and signal‐to‐noise ratio identifies potential impacts to image quality and dose burdens to patients. The results highlight the need for careful consideration when using the AEC and the requirement for accurate positioning technique by the radiographer.

This study highlights the impact of inaccurate patient positioning on dose and image quality when utilising the automatic exposure control (AEC) for radiographic examinations. While the AEC serves to optimise radiation dose to patients, the results here demonstrate that poor positioning can lead to significant dose implications for patients.

1. Introduction

Automatic exposure control (AEC) is a device used in radiographic practice which terminates a radiographic acquisition when a certain exposure threshold is reached [1, 2]. This technology aims to optimise image quality while reducing radiation dose to patients, irrespective of their body habitus or body region being examined [2, 3, 4]. Several ionisation chambers, typically 3 to 5, are situated inside the bucky and in front of the image receptor. These chambers detect the number of photons remaining after the beam passes through the patient's anatomy [1, 2, 5, 6, 7]. These photons are collected and their signal amplified until a certain, preset point is reached, at which a break in the system occurs and the exposure times out or is completed [1, 2, 5, 6, 7]. The desired level of radiation is predetermined by the manufacturer or institution's medical physicist to ensure image quality does not hinder a radiologist's reporting confidence [8].

The AEC is commonly utilised in clinical practice, for in‐bucky imaging including but not limited to the chest, abdomen, pelvis and the spine [4, 8, 9, 10, 11]. These regions of the body typically have the greatest variations in patient size and density and, combined, contain many radiosensitive organs [9, 10]. These examinations also have the highest radiation burden of any areas of the body and are the most susceptible to positioning variation [12, 13]. If used correctly, AEC can reduce repeat rates, reduce patient dose and increase the speed at which patients can be examined [5, 14]. When positioned correctly, the AEC allows for replication of the same or similar examinations repeatedly, with very different patient thicknesses and anatomy, theoretically removing the possibility of exposure selection errors by radiographers [4, 6, 8].

Where variability in patient positioning leads to misuse of the AEC, reduced patient dose may lead to underexposure. This may compromise image quality, leading to repeat examinations, increasing patient dose and impacting resource utilisation. This study aimed to evaluate the changes, if any, to entrance skin dose (ESD) and image quality resulting from incorrect patient positioning technique. The study sought to observe changes when using AEC for chest, abdomen, lumbar spine and pelvic x‐ray examinations. The overarching research question being, how do changes in patient positioning (from centred to off‐centred) affect the patient's ESD and image quality?

The optimisation of radiographic exposure, and by extension, patient dose, with the use of the AEC, assumes accurate measures are in place. This includes correct patient positioning and chamber selection, among other factors. Careful positioning considerations and alignment must be followed when using the AEC, to ensure dose optimisation can occur to an acceptable standard. When using the AEC, it can be difficult for areas of interest to be positioned correctly over the necessary chamber due to variation in anatomy, hard‐to‐locate regions of interest or user error [12]. When the position of the chamber and the position of the patient do not accurately coincide, the AEC will time out incorrectly, causing an over or under exposure of the subsequent image [5, 6, 15, 16]. For example, studies have assessed the relationship of pelvis positioning and AEC use, typically suggesting a reduction in dose of 31% and 36.8% and altered image quality when positioning patients with the patient's head at opposite ends of the table and therefore assuming differing anatomical structures present over the ionisation chambers [12, 17]. Other studies have investigated dose discrepancies when utilising differing AEC chambers and found over 40% more dose when using one chamber against another [5, 11, 12]. Kim et al. also noted that there was no ‘statistically meaningful difference’ between the multitude of chamber combinations when reviewing the image quality in their study [11].

AEC chamber selection is also of vital importance to ensure dose is kept at a minimum [2, 5, 10, 12]. Some examinations will benefit from the use of multiple chamber selection, with exposure cut‐off when an average dose is met across the chambers, and others where a single chamber is utilised and positioned over a specific anatomical region. This has led to common practice for the use of two lateral chambers for PA chest and pelvis imaging, the use of a single, central chamber for AP lumbar spine, and all three chambers for AP abdominal imaging.

Early studies demonstrated that utilising the AEC could reduce operator‐dependent exposure errors compared to manual technique and the consideration for AEC use as a dose reduction tool [1, 2, 9, 18, 19]. Anecdotally, the advent of digital radiography has reduced an onus on exposure selection, and understanding of exposure parameters has relaxed. Modern radiographers tend to be more reliant on AEC for exposure selection, particularly for examinations that typically have a wide variation in patient size.

There are many benefits to utilising the AEC in general radiographic practice when awareness and knowledge of the technology are thoroughly understood. Many studies describe that incorrect and complacent implementation of automatic exposure techniques can cause greater harm to patients than advantage [6, 14, 20, 21]. Search of the available literature suggests large discrepancies in dose modulation for routine examinations as well as under or over exposure of images, attributed to misuse of the AEC [15, 22, 23]. These problems most commonly arose from incorrect positioning, combined with poor chamber selection and a lack of understanding of how the AEC produces a diagnostic image [5, 10, 12, 15, 17]. To the contrary, when patients were positioned correctly, adhering to common positioning considerations, the AEC worked to reduce repeat rates and ensure the ALARA principle was maintained in the clinical space, reducing dose to patients and preventing dose creep [9, 18, 19, 24, 25]. Choosing the correct ionising chambers and aligning anatomical regions of interest with these chambers should optimise image quality [5, 12].

2. Methods

An ALDERSON anthropomorphic phantom was marked, under fluoroscopic conditions to ensure no parallax error, in 21 positions for each anatomical area. Each mark corresponded to the location of the central ray for use in the study. The first mark located the ideal central ray position based on department protocols and commonly accepted points. Five points were then marked 1 cm from the ideal in each direction: superior, inferior, left lateral and right lateral. The central ray was then aligned to each mark for exposure.

A General Electric (GE) Definium 656 HD X‐Ray System with integrated 41 × 41 cm flat‐panel detector (GE Healthcare, Chicago, USA) was utilised. All images were performed using the AEC, collimated to 41 × 41 cm and with the central ray perpendicular to the centre of the appropriate wall or table bucky. A source to image distance of 100 cm for the pelvis and lumbar spine, 120 cm for the abdomen and 180 cm for the chest was implemented. A 12:1 grid for the table bucky (used for the abdomen, pelvis and lumbar spine images) and a 13:1 grid for the wall bucky (used for the chest images) was used to simulate clinical conditions. Tube potential of 80 kVp for the abdomen, pelvis and lumbar spine images and 120 kVp for the PA chest were used, mimicking the clinical conditions of the organisation and vendor recommendations. Likewise, chamber selection was based on the department's preset protocols, as arranged with the vendor, with the single middle chamber selected for the AP lumbar spine, all chambers selected for the AP abdomen, and the two lateral chambers selected for the PA chest and AP Pelvis images.

The air kerma was measured using a dosimeter for each image, placed on the phantom within the X‐ray field and in an identical position across all images that also avoided superimposition of the ionisation chambers. The phantom was initially exposed 10 times in the ideal central position to test for air kerma variability. The results were consistent, and so each position was then exposed five times thereafter. Note is made that the first five results in the ideally centred position were used for statistical analysis. Figure 1 identifies the ideal centring points for the four body areas, and Figure 2 demonstrates an example of the phantom marked for use.

FIGURE 1.

Diagram showing the location of ideal centring points used for study.

FIGURE 2.

Example of preparation of marked‐up phantom for use for study.

Measurements for mAs and air kerma were recorded for each image, as were three separate regions of interest (ROI) measurements (measuring 0.5 mm for all measurements), which then recorded the mean value, standard deviation and signal‐to‐noise (SNR) ratio for the image of each off‐set that recorded the lowest dose. The position of the ROIs is provided in Table 1. Differences in mean dose values from off‐centred to centred images were then expressed and tabled as a percentage.

TABLE 1.

Position of region of interest (ROI) measurements.

	Position of ROI
AP Chest	Over spinous process of 7th thoracic vertebrae
	Over left lung parenchyma in centre of chest between 4th and 5th Rib
	Over heart between 7th and 8th Rib
AP Supine Abdomen	Over spinous process of 3rd lumbar vertebrae
	Over right ala measured 3 cm inferior to iliac crest
	Over right abdominal soft tissue 5 cm lateral to the 2nd lumbar spinous process
AP Pelvis	Over right ala measured 3 cm inferior to iliac crest
	Over centre of 1st sacral segment 2 cm inferior to L5/S1 joint space
	Over pubic bone immediately to the left of pubic symphysis
AP Lumbar Spine	Over spinous process of 3rd lumbar vertebrae
	Over right ala measured 3 cm inferior to iliac crest
	Over right abdominal soft tissue 5 cm lateral to the 2nd lumbar spinous process

3. Statistical Analyses

A multiple linear regression was performed using IBM SPSS Statistics (v28.0.1.0, NY, USA) using distance from centre (0 cm) in the lateral and vertical directions as predictors for dosage in each ROI. A two‐way multivariate analysis of variance (MANOVA) was performed to determine the distance threshold from the centre at which dosage is significantly affected within each ROI. Data were plotted as contour heatmaps generated in Matlab (R2024b, v24.2.0.2712019, MA, USA).

Differences in SNR values from off‐centred to centred images were then expressed and tabled as a percentage.

4. Results

A total of 440 radiographs were performed, corresponding to 440 measurements for mAs and air kerma and 252 ROI, standard deviation and SNR measurements.

Air‐kerma changes recorded a maximum reduced air kerma of 76.07%, noted for the AP pelvis offset laterally to the left by 5 cm (see Table 3) to maximum increased air kerma of 50.09% where the AP chest was offset to the right by 4 cm (also noted at Table 3). The greatest difference in air‐kerma range was seen for pelvic imaging which had a variation in air kerma of 84.07%. The smallest difference in air‐kerma range was seen for lumbar imaging which had a variation in air kerma of 45.98%. The mean mAs and air kerma were also tabled and expressed as a percentage and are presented in Table 2.

TABLE 3.

Signal to noise ratio (SNR) measurements recorded for the observation of each off‐set that recorded the lowest dose as well as the percentage of change in SNR from the control (or ideally centred) value.

AP chest								Supine abdomen
Body region/controls		T7	Change to SNR (%)	Lung fields (b/w ribs 4–5)	Change to SNR (%)	Heart	Change to SNR (%)	Body region/controls		L3	Change to SNR (%)	Right Ala	Change to SNR (%)	Left soft tissue	Change to SNR (%)
Control	180 cm							Control	120 cm
	120kVp	65.46		22.20		63.82			80kVp	19.02		84.75		74.68
Superior	1	67.52	−3.15	22.49	1.3	67.21	5.31	Superior	1	19.81	4.15	88.52	4.45	73.02	−2.22
	2	62.77	−4.11	21.36	−3.78	67.64	5.98		2	20.17	6.05	89.52	5.63	71.75	−3.92
	3	57.87	−11.60	20.33	−8.42	60.63	−5		3	20.99	10.36	85.61	1.01	76.70	2.7
	4	54.86	−16.19	19.90	−10.36	57.82	−9.4		4	20.98	10.3	78.27	−7.65	73.94	−0.1
	5	51.87	−20.76	19.27	−13.2	62.99	−1.3		5	21.31	12.04	76.38	−9.88	72.25	−3.25
Inferior	1	70.47	7.65	22.22	0	73.78	15.6	Inferior	1	19.73	3.73	90.90	7.26	78.88	5.62
	2	73.49	12.27	20.54	−7.48	52.88	−17.14		2	18.95	−0.37	84.15	−0.71	72.37	−3.09
	3	69.21	5.73	20.04	−9.73	52.08	−18.39		3	19.13	0.58	88.77	4.74	70.15	−6.06
	4	66.16	1.07	19.81	−10.76	53.34	−16.42		4	18.84	−0.95	88.43	4.34	66.85	−10.48
	5	62.52	−4.49	19.95	−10.13	47.45	−25.65		5	18.16	−4.52	87.07	2.74	67.04	−10.23
Left lateral	1	48.70	−25.60	23.21	4.55	47.53	−25.52	Left lateral	1	19.55	2.78	82.92	−2.16	70.50	−5.6
	2	56.43	−13.79	24.78	11.62	52.04	−18.46		2	19.62	3.15	82.91	−2.17	73.62	−1.42
	3	47.28	−27.77	16.99	−23.47	37.69	−40.94		3	18.52	−2.63	74.72	−11.83	67.79	−9.23
	4	56.32	−13.96	25.48	14.77	32.46	−49.14		4	18.38	−3.36	70.84	−16.41	59.17	−20.77
	5	46.59	−28.96	24.59	10.76	34.77	−45.52		5	19.66	3.36	67.71	−20.11	56.02	−25
Right lateral	1	61.28	−6.39	24.99	12.57	56.88	−10.87	Right lateral	1	20.32	6.83	83.79	−1.13	76.97	3.07
	2	56.23	−14.10	22.77	2.57	66.74	4.57		2	19.94	4.84	77.51	−8.54	69.60	−6.8
	3	49.40	−25.43	21.62	−2.61	52.33	−18		3	18.88	−0.73	78.58	−7.28	76.78	2.81
	4	46.64	−28.75	22.60	1.8	50.38	−21.06		4	18.48	−2.84	69.52	−17.97	77.27	3.47
	5	51.19	−21.80	22.28	0.36	54.47	−14.65		5	18.48	−2.84	66.43	−21.62	70.64	−5.41

Supine pelvis								Supine lumbar spine
Body region/controls		Right Ala	Change to SNR (%)	S1	Change to SNR (%)	Pubis	Change to SNR (%)	Body region/controls		L3	Change to SNR (%)	Right Ala	Change to SNR (%)	Left soft tissue	Change to SNR (%)
Control	100 cm							Control	100 cm
	85kVp	34.68		14.84		47.17			80kVp	15.26		98.83		67.47
Superior	1	33.16	−4.38	14.62	−1.48	48.25	2.29	Superior	1	16.11	5.57	96.56	−2.3	62.38	−7.54
	2	33.57	−3.2	13.95	−6	45.78	−2.95		2	17.06	11.79	95.62	−3.25	67.47	0
	3	30.66	−11.59	13.67	−7.88	39.49	−16.28		3	17.98	17.82	100.23	1.42	70.37	4.3
	4	28.27	−18.48	12.52	−15.63	37.72	−20.03		4	18.64	22.15	97.59	−1.25	73.53	9
	5	28.63	−17.44	12.41	−16.37	40.14	−14.84		5	19.09	25.1	109.67	10.97	67.34	−0.19
Inferior	1	37.71	8.74	15.17	2.22	50.60	7.27	Inferior	1	15.28	0.13	99.30	0.47	65.84	−2.41
	2	35.23	1.59	14.91	0.47	48.06	1.89		2	15.15	−0.72	95.33	−3.54	63.06	−6.54
	3	33.72	−2.77	14.93	0.6	49.92	5.83		3	14.85	−2.69	90.02	−8.91	57.66	−14.54
	4	35.53	−2.45	14.70	−0.94	49.21	4.32		4	14.83	−2.82	71.84	−27.31	55.04	−18.42
	5	32.94	−5.02	15.65	5.46	50.48	7.02		5	15.04	−1.44	74.81	−24.3	52.11	−22.76
Left lateral	1	32.39	−6.6	14.73	−0.74	46.84	−0.7	Left lateral	1	14.85	−2.69	98.01	−0.83	63.69	−5.6
	2	32.53	−6.2	13.71	−7.61	44.09	−6.53		2	15.71	2.95	96.97	−1.88	63.60	−5.73
	3	34.26	−1.21	13.24	−10.78	41.68	−11.64		3	16.79	10.03	85.93	−13.05	56.61	−16.1
	4	32.03	−7.64	12.55	−15.43	37.98	−19.48		4	16.70	9.44	79.65	−19.41	50.37	−25.34
	5	27.70	−20.13	17.73	19.47	31.15	−33.96		5	16.61	8.85	82.67	−16.35	56.63	−16.1
Right lateral	1	34.39	−0.84	15.89	7.07	55.23	17.09	Right lateral	1	15.57	2.03	106.18	7.44	60.06	−11
	2	35.29	1.76	17.54	18.19	53.13	12.63		2	15.11	−0.98	96.94	−1.91	56.27	−16.6
	3	35.62	2.71	17.04	14.82	49.89	5.77		3	13.94	−8.65	89.81	−9.13	72.00	6.71
	4	32.35	−6.72	18.33	23.52	48.05	1.86		4	12.98	−14.94	64.60	−34.63	62.28	−7.7
	5	25.67	−26	17.93	20.82	47.24	0.15		5	12.93	−15.27	60.77	−38.51	55.84	−17.24

Note: Note the maximum change in SNR highlighted in yellow.

TABLE 2.

Mean mAs and air‐kerma results for off‐set positioning, including the percentage change for all observations.

AP chest						Supine abdomen
Body region/controls		Mean mAs	Change in mAs (%)	Mean air kerma (μGy)	Change in AK (%)	Body region/controls		Mean mAs	Change in mAs (%)	Mean air kerma (mGy)	Change in AK (%)
Control	180 cm					Control	120 cm
	120kVp	1.10		56.57			80kVp	18.19		1.28
Superior	1	1.12	1.80	57.50	1.65	Superior	1	17.12	−5.88	1.21	−5.33
	2	1.15	4.54	59.64	5.43		2	16.45	−9.57	1.16	−9.10
	3	1.19	8.18	61.80	9.24		3	15.81	−13.08	1.12	−12.72
	4	1.18	7.27	61.26	8.28		4	14.84	−18.42	1.04	−18.34
	5	1.04	−5.45	53.14	−6.06		5	13.88	−23.69	0.97	−24.05
Inferior	1	1.07	−2.72	55.03	−2.72	Inferior	1	17.76	−2.36	1.25	−2.45
	2	1.07	−2.72	54.87	−3.01		2	18.10	−0.49	1.27	−1.01
	3	1.08	−1.82	55.59	−1.74		3	18.14	−0.27	1.26	−1.43
	4	1.09	−0.91	55.63	−1.67		4	18.25	0.33	1.26	−1.49
	5	1.12	1.80	56.25	−0.57		5	18.64	2.47	1.28	−0.16
Left lateral	1	1.10	0.00	56.03	−0.96	Left lateral	1	17.13	−5.83	1.20	−6.33
	2	1.13	2.73	57.50	1.64		2	15.39	−15.39	1.07	−16.60
	3	1.27	15.45	64.29	13.64		3	12.66	−30.40	0.87	−32.23
	4	1.41	28.18	71.67	26.69		4	9.46	−47.99	0.64	−49.90
	5	1.49	35.45	75.64	33.71		5	4.90	−73.06	0.32	−74.72
Right lateral	1	1.15	4.54	58.68	3.72	Right lateral	1	16.98	−6.65	1.21	−5.59
	2	1.34	21.82	68.07	20.33		2	15.91	−1.25	1.14	−10.97
	3	1.55	40.91	79.20	40.00		3	14.70	−19.18	1.06	−17.31
	4	1.71	55.45	87.17	50.09		4	13.14	−27.76	0.88	−31.29
	5	1.60	45.45	81.39	43.87		5	10.35	−43.10	0.75	−41.22

Supine pelvis						AP lumbar spine
Body region/controls		Mean mAs	Change in mAs (%)	Mean air kerma (μGy)	Change in AK (%)	Body region/controls		Mean mAs	Change in mAs (%)	Mean air kerma (μGy)	Change in AK (%)
Control	100 cm					Control	100 cm
	85kVp	5.85		731.08			80kVp	8.50		930.36
Superior	1	5.66	−3.25	713.70	−2.38	Superior	1	8.35	−1.76	914.46	−1.71
	2	5.69	−2.74	721.38	−1.33		2	8.04	−5.41	878.18	−5.61
	3	5.54	−5.30	706.84	−3.32		3	7.73	−9.06	834.72	−10.28
	4	5.50	−5.98	704.74	−3.60		4	7.32	−13.88	783.46	−15.79
	5	5.37	−8.21	689.66	−5.67		5	7.11	−16.35	756.74	−18.66
Inferior	1	5.99	2.39	739.34	1.13	Inferior	1	8.74	2.82	957.14	2.88
	2	6.09	4.10	738.50	1.01		2	9.04	6.36	985.52	5.93
	3	6.22	6.32	745.42	1.96		3	9.15	7.65	990.80	6.50
	4	6.44	10.09	759.60	3.90		4	9.55	12.35	1026.40	10.32
	5	6.80	16.24	789.56	8.00		5	9.57	12.59	1019.48	9.58
Left lateral	1	5.44	−7.01	675.90	−7.55	Left lateral	1	8.56	0.71	926.94	−0.37
	2	4.93	−15.73	597.24	−18.31		2	8.31	−2.24	878.12	−5.62
	3	3.54	−39.49	423.48	−42.07		3	7.64	−10.12	798.76	−14.15
	4	2.05	−64.96	240.38	−67.12		4	6.83	−19.65	701.16	−24.64
	5	1.52	−74.02	174.92	−76.07		5	6.26	−26.35	598.58	−35.66
Right lateral	1	5.98	2.22	753.84	3.11	Right lateral	1	8.69	2.24	965.68	3.80
	2	5.64	−3.59	717.50	−1.86		2	8.65	1.76	970.18	4.28
	3	5.36	−8.38	686.20	−6.14		3	8.42	−0.94	953.74	2.51
	4	4.32	−26.15	548.48	−24.98		4	8.41	−1.06	960.04	3.19
	5	2.62	−55.21	333.26	−54.42		5	7.86	−7.53	896.88	3.60

Note: Note is made of the maximum mAs and air‐kerma changes highlighted in yellow and also the greatest range of difference in air‐kerma and mAs observed in the pelvis of 84.07% and 90.26% respectively, between off‐centring of 5 cm to the right and 5 cm inferiorly.

5. Statistical Analyses

The results of the multiple linear regression revealed that dosage significantly differs as distance from the centre increases in the AP chest (F = 260.49, p < 0.001, Adj. R² = 0.90), supine abdomen (F = 380.44, p < 0.001, Adj. R² = 0.93), supine pelvis (F = 308.34, p < 0.001, Adj. R² = 0.91) and supine lumbar spine (F = 278.83, p < 0.001, Adj. R² = 0.90).

For the AP chest region, the relationship between dosage and distance from centre is represented by the following regression equation:

$$\begin{array}{c}\mathit{\text{Dosage}}=1.081+0.074\left(\mathit{\text{Left lateral}}\right)+0.004\left(\mathit{\text{Inferior}}\right)\hfill \\ +0.129\left(\mathit{\text{Right lateral}}\right)+0.013\left(\mathit{\text{Superior}}\right)\hfill \end{array}$$

In the AP chest region, movement in the lateral direction away from centre is the most significant predictor of increase in dosage (with right being a stronger predictor than left, p < 0.001). Changes to dosage in the inferior direction are not significant, while in the inferior direction, a small but statistically significant positive effect on dosage was found (p = 0.004, Figure 3).

FIGURE 3.

Heat map of dose change for the AP Chest (a), AP Abdomen (b), AP Pelvis (c) and AP Lumbar Spine (d), demonstrating dose changes as positioning migrates from ideal centre. Indications for the off‐centring recordings note the statistically significant changes (p < 0.05) to dose (red) and recordings that do not observe statistical significance (green).

For the supine abdomen region, the relationship between dosage and distance from centre is represented by the following regression equation:

$$\begin{array}{c}\mathit{\text{Dosage}}=18.766-2.423\left(\mathit{\text{Left lateral}}\right)-0.126\left(\mathit{\text{Inferior}}\right)\hfill \\ -1.533\left(\mathit{\text{Right lateral}}\right)-1.005\left(\mathit{\text{Superior}}\right)\hfill \end{array}$$

Similar to the AP chest, in the supine abdomen region, movement in the lateral direction away from centre is the most significant predictor of decrease in dosage (with left being a stronger predictor than right, p < 0.001). Movement in the superior direction was also significantly associated with decrease in dosage (p < 0.001), while movement in the inferior direction did not significantly affect dosage (Figure 3).

For the supine pelvis region, the relationship between dosage and distance from centre is represented by the following regression equation:

$$\begin{array}{cc}\mathit{\text{Dosage}}=\hfill & 6.040-0.888\left(\mathit{\text{Left lateral}}\right)+0.109\left(\mathit{\text{Inferior}}\right)\hfill \\ \hfill & -0.488\left(\mathit{\text{Right lateral}}\right)-0.147\left(\mathit{\text{Superior}}\right)\hfill \end{array}$$

While changes in dosage were lower in magnitude in the supine pelvis region compared to the other ROIs, movement in all directions away from centre was associated with small, but significant changes in dosage (p < 0.001, Figure 3).

Finally, for the supine lumbar spine region, the relationship between dosage and distance from centre is represented by the following regression equation:

$$\begin{array}{c}\mathit{\text{Dosage}}=8.714-0.437\left(\mathit{\text{Left lateral}}\right)+0.175\left(\mathit{\text{Inferior}}\right)\hfill \\ -0.119\left(\mathit{\text{Right lateral}}\right)-0.333\left(\mathit{\text{Superior}}\right)\hfill \end{array}$$

Similar to the supine pelvis region, in the supine lumbar spine region, movement in all directions away from centre was associated with a small, but significant change in dosage (p < 0.001, Figure 3).

Two‐way MANOVA found that in the AP chest region, changes to dosage were significant after 2 cm from centre in the superior and left lateral directions, 1 cm in the right lateral direction and not significantly different for any distance from centre in the inferior direction. Similarly, in the supine abdomen region, changes to dosage were significant at even 1 cm away from centre in all directions except for inferior, which did not significantly impact dosage, even as far as 5 cm off centre.

In the supine pelvis, changes to dosage were significantly impacted at 4 cm away from centre in the superior direction and 2 cm away from centre in the inferior direction. The supine pelvis was particularly susceptible in the left lateral direction, significantly impacted immediately off centre at 1 cm and from 3 cm in the right lateral direction.

Finally, in the supine lumbar spine, dosage was significantly impacted in both the superior and inferior directions at 2 cm away from centre, 3 cm in the left lateral direction and 5 cm in the right lateral direction.

Changes in air kerma were statistically significant in 54 of the 80 (67.50%) observed off‐centred positions.

Changes to SNR varied from a maximum negative change of 49.14% to a maximum positive change of 25.10%. Changes to SNR greater than 10% were demonstrated in 38.33% (92/240) of positional changes. Changes greater than 10% to SNR were recorded more frequently for lateral off‐centring (56 of the 92 or 60.87%) of observations that were greater than 10%, especially left lateral (32/92 or 34.78%) and least frequently for inferior off‐centring (15/92 or 16.3%) of occurrences where changes to SNR were observed greater than 10%. Table 3 demonstrates the recorded SNR values as well as the percentage difference in SNR values from centred to off‐centred positions.

Changes in SNR were observed to be variable, and no pattern was noted for changes in SNR in relation to air‐kerma changes and associated off‐centring.

6. Discussion

Air kerma is essentially a representation of the value, or number, of x‐ray photons present within the x‐ray field and incident on the detector and is often loosely interchanged as a measurement of radiation dose. For this reason, the range of changes to the air kerma (or dose) as a result of mispositioning demonstrated in this study reads as concerning for radiographers unskilled in accurate patient positioning techniques and over‐reliant on AEC for exposure selection.

This study has focused on percentage changes to dose rather than the numerical values for mAs and air kerma. This was chosen because this is a phantom study, which will considerably undervalue the results for mAs and air kerma against real‐life values. Changes to mAs for abdomen and pelvis imaging demonstrated variance in recorded measurements by up to 4.5 times a maximum to minimum recording. As an example, note the minimum and maximum recorded mAs values for the pelvis in this study of 1.52 mAs and 6.80 mAs, attributed solely to mispositioning of the same patient. Clinical context for such an observation would be that an exposure of around 20 mAs, somewhat routine for the pelvis, would infer a difference from this exposure of close to 100 mAs, or conversely 4 mAs, which is alarming. This paints a picture of the impact that poor positioning may have on patient safety, confidence in dose and image quality, and adherence to the ALARA principle.

Body habitus variability that exists across patients further exacerbates these concerns, given that the changes seen may be greater for larger patients. Also, larger patients tend to be more difficult to position accurately given the difficulty of accurately palpating bony landmarks, vital for precise positioning and therefore more likely to incur the larger variations observed in this study.

Dose changes were more commonly noted to be statistically significant if off‐centring was greater than 1–2 cm from the ideal centring point. Also, greater dose changes were more commonly observed in lateral displacements, which would potentially correlate with the ionising chambers aligning over differing tissue types, which may attenuate radiation differently. When solid bony anatomy overlies the AEC chamber, the exposure will be significantly prolonged by the AEC as there is greater attenuation of the beam, compared to if it were an area of soft tissue in the same position and vice versa [20]. Anecdotally, this is more likely to occur with larger patients, and our results suggest a likely greater dose burden and exposure‐related error for these patients with over‐reliance on the AEC and inexact positioning.

Statistically significant changes in dose were observed with < 1 cm of mispositioning for the chest, abdomen and pelvis, and < 2 cm for the lumbar spine depending on the direction of off‐centring. Understanding chamber position is vital to safe radiation practice and is supported by the results here. As an example, when performing a ‘Charnley’ pelvis (an orthopaedic projection that centres on the pubic symphysis rather than the midpoint of the anterior superior iliac spine), one should question the appropriateness of the use of the AEC given the position of the chambers in relation to anatomy, which will have differing tissues and attenuation coefficient to the standard pelvis position. Furthermore, our results would suggest that the use of the AEC be contraindicated when positioning a pelvis centred on the pubic symphysis, with a noted statistically significant change to dose. This supports and extends the position proposed in previous studies, where positioning an AP pelvis upside down on the table (i.e., with the feet to the pre‐set head end of the examination table) infers mis‐positioning of the appropriate anatomy over the AEC chambers and leads to a greater than 30% change in dose [12, 17].

Similarly, some of the largest over‐irradiation changes in dose occurred where the chest was off‐centred laterally, especially to the right. This could be explained by the ionisation chambers in such an offset being positioned over the right shoulder and partly over the mediastinum, rather than the air‐filled apices, which would have a much lower attenuation of the radiation. Statistically significant changes to dose were noted in almost all off‐centred observations for the chest, with the exception of inferior movements. This highlights the variability in dose that occurs for a body area that is commonly imaged using the support of the AEC.

The results for the AP abdomen mirrored that of the chest but showed that any mispositioning beyond perfect centring, except for inferior off‐centring, would result in statistically significant changes in dose. Interestingly, however, the abdomen yielded the least variance in SNR changes. This might possibly be due to the use of three AEC chambers averaging the accepted dose and, therefore, allowing for less volatility to changes in dose.

The pelvis and spine demonstrated slightly less statistically significant recordings, but observed some of the greater changes in dose, particularly when off‐centred to the left. These results are a little difficult to explain, given what should be the symmetry of the anatomy. Regardless, the results still validate the likelihood of severe dose implications with relatively minimal off‐centring (> 1–2 cm).

Image quality was difficult to assess given the use of the phantom rather than human tissue and not extensively explored here; however, SNR values varied by close to 75%, suggesting potentially considerable impacts to image quality. Large changes in SNR were more consistently recorded for lateral displacements in centring, mirroring that seen for dose results and explained similarly. Close to 40% of measurements resulted in changes to SNR of greater than 10%, which might suggest genuine impact to image quality in the clinical environment and which further highlight and support the need for accurate patient positioning when utilising the AEC. This is an area for further research and exploration; however, the somewhat variable results observed to changes in SNR do tend to support observations by Kim et al. that note no ‘statistically meaningful difference’ observed when assessing variability introduced through the use of the AEC [11].

It could be reasoned from these results that poor patient positioning could likely result in an overexposure to patients, due either to the AEC seeking a greater than required exposure or under‐exposing by such a degree as to render the image quality unacceptable and a repeat being required. Regardless, our results attest to the requirement for impeccable patient positioning techniques and a comprehensive understanding of the AEC system and chamber position as well as exposure selection technique. Our results demonstrate that utilising the AEC with accurate positioning still has strong advantages to the clinical radiographer; however, it also emphasises real flaws with the system if inaccurate positioning is not achieved. The results suggest that careful consideration of the use of AEC should be undertaken by the clinical radiographer and warn of over‐reliance on the system.

7. Limitations

The phantom used in this study, while designed to mimic human density and attenuation, was old, and the thickness and composition of it were not likely representative of human tissue. Therefore, the radiographs did not adequately represent the imaging characteristics of patients, likely under‐representing real life values. Also, in clinical practice, collimation for the lumbar spine would be < 41 cm, which would also affect the image quality and dose. Quality assurance measures were performed on the radiographic equipment, including the AEC, but not the dosimeter, potentially also contributing to measurement bias, which may overestimate or underestimate the true dose. Given that the dosimeter used was the same for all recordings and occurred on the same day, it is reasonable to accept that the changes noted in the study are a true representation for comparison.

Further limitations include the fact that this was a single equipment model, as well as a phantom‐only design, which therefore assumes clinical validation. The study also utilised a limited range of exposure settings, including collimated field size, which present as points for investigation in any future studies. Imaging, recording and data collection were all completed by the same two operators, which may introduce inter‐operator variability.

8. Conclusion

This study highlights the importance of patient positioning on dose and image quality when utilising the AEC. Statistically significant differences in dose were commonly discovered for mis‐positioning beyond 1–2 cm, which can and will occur within the clinical environment, particularly for larger and more complicated patients. SNR changes also varied greatly, suggesting considerable effects on image quality.

The greatest difference in air‐kerma range was seen for pelvic imaging, which had a variation in air kerma of 84.07%, and the smallest difference in air‐kerma range was seen for lumbar imaging, which had a variation in air kerma of 45.98%. Relating these findings to the clinical environment, these results suggest likely excursions in exposure, which will have a profound impact on patient dose, image quality, confidence in exposure selection, possible repeat imaging and lack of adherence to the ALARA principles.

The results highlight the need for careful consideration of the use of AEC by the radiographer with consideration for the accuracy of patient positioning in relation to the position of the AEC chambers. The study highlights the need for radiographers to be conscious of precise positioning techniques, utilising positioning that accurately identifies accepted central ray alignment for the body region to be imaged.

This study utilised a single equipment model, a limited range of exposure settings and data collection by the same operators but still highlights the variability that is exposed when utilising the AEC inappropriately. New technologies, such as automatic centring of patients using cameras affixed to the X‐ray tube, may alleviate some of these errors but must also be investigated for consistency and precision of positioning.

Author Contributions

All authors contributed significantly to this project and are in agreement over the content of the manuscript.

Ethics Statement

Ethical approval was sought and granted for this retrospective study by The Western Health Low Risk Ethics Panel on 17th October 2023.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: jmrs70030‐sup‐0001‐DataS1.xlsx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.