Evaluation begins with high-quality plain radiographs, especially post reduction. Essential projections include: a true antero-posterior view in the scapular plane (Grashey view) with the arm in maximal internal-rotation, axillary view and Y view (less important). X-rays allow the identification also of associated fractures, which are more frequent in older patients, such as: tuberosities, surgical or anatomical humeral neck and coracoid (rare). Several other more specific projections could better detect and characterize the antero-inferior profile of the glenoid fossa: the West Point view (a particular kind of axillary view) and the Bernageau view.

Itoi, et al. [23] reported that the West Point view has shown a high correlation with computed tomography (CT) imaging in estimating glenoid bone loss.

Some authors have also reported the Bernageau view as a valid and accurate method for detecting antero-inferior glenoid bone loss: Edwards, et al. [24] reported that the Bernageau method can identify a glenoid rim defect in 79% of shoulders with anterior instability; Griffith, et al. [25, 26] found that images obtained with the Bernageau view are similar to axial images of 2-dimensional CT ; Pansard, et al. [27] state that glenoid bone defect measurement via the Bernageau profile view is a valid and reliable method that could decrease CT indications. To reduce the radioactive exposure of the Bernageau method, Sugaya has proposed a modified Bernageau method whereby patients lie on their axilla in the so called “TV watching position” [28].

Gerber, et al. [29] published that the true antero-posterior and Bernageau views detect bone loss of the antero-inferior glenoid profile with a sensitivity of 66% and specificity of 100%: however the limitation is the low reliability regarding the defect size; for detection of humeral head defects the Striker Notch view could be helpful, as it can better delineate their size and orientation; Bushnell, et al. [21] reported that only 57% of operative bony defects were detected on pre-operative radiographs.

Therefore CT imaging is required to obtain more accurate information each time a bone defect is identified on plain radiographs, or the clinical findings suggest their presence. CT scanning is the method of choice for studying glenoid bone loss [18, 25, 26, 30] and for humeral head defects, albeit with less sensitivity; however, there are few studies specifically examining CT scanning on Hill- Sachs lesions [20] compared to glenoid defects [23, 25, 30-33].

Following the detection of bone defects, the role of imaging shifts to their characterization and quantification for the purposes of treatment choice and planning. Depending on the severity of bone loss versus soft tissue damage, the surgeon should decide between the open or arthroscopic approaches.

Fig. (4a). A case of bilateral engaging Hill-Sachs lesion on AP X-ray 108x58mm (150 x 150 DPI).

Fig. (4b). CT axial view with visualization of bilateral engaging wide and deep Hill-Sachs lesion 89x60mm (150 x 150 DPI) .

Fig. (4c). CT coronal view of the same case 111x44mm (150 x 150 DPI).

CT scanning must answer some precise questions about bone defects: presence or absence, location, morphology and size.

For glenoid defects a location description can be expressed in relation to the glenoid clock face, typically they are situated on the antero-inferior rim, mainly anterior [25, 34]. Moreover, they can present morphologically with erosion (attritional bone loss) or fracture [18, 25, 26, 35]. Bigliani, et al. [36] coined the term bony-Bankart lesion and classified the glenoid bone defects into three types with increasing severity: the third type is bone loss: < 25% (3-A) and > 25% (3-B). Edwards, et al. [24] classified the glenoid defects into three types: bony-Bankart; loss of the antero-inferior angle with no visible fragment (cliff-sign) and bone loss (blunted angle). Griffith, et al. [26] also categorized into three types: the classic bony-Bankart lesion; “the straight pear” lesion, with a bone loss of 5-15% (the more frequent type in recurrent instability) and “the inverted pear” lesion, more severe as presenting with a bone loss > 20%. There is no consensus on the critical value for defining as severe or at high risk of recurrence after surgical repair without bone grafting, but most authors agree that the limit is about 20-25% of the total glenoid surface [16, 23, 30, 36-39].

Fig. (5). Measure of the lenght of Hill-Sachs lesion on CT sagittal view 55x46mm (150 x 150 DPI).

The relationship between glenoid bone defect and recurrence is known. Rowe, et al. [40] reported a recurrence rate of 6-67% depending on the type of bone defect; Boileau, et al. [41] found a 75% recurrence rate among patients with at least 25% glenoid bone width loss; Burkhart, et al. [37, 39] found a recurrence rate of 89% in patients with 25-27% loss of glenoid width (“inverted pear” glenoid). This exemplifies how the real problem is quantifying the glenoid bone loss. Several methods have been used in CT to calculate the defect, grouped into linear (Bankart length measurement [29], Ratio method [42]) and circular (Best-fit or True circle method [28], Superimposed circle method [30], PICO method [32]), so the bone loss is calculated in mm or mm2 and expressed as the percentage of length or area of bone loss versus the total glenoid fossa respectively. Some methods calculate based on 2D-CT data with or without MPR (multiplanar reconstruction), others based on 3D-CT; some methods have the limitation of using the contralateral shoulder (if intact) as the control, but do gain in accuracy. The glenoid is always studied on the “en face view” which shows it in its entirety, with the exception of the humeral head.

Most methods are based on the principle that the inferior part of the glenoid has a circular shape [43], attention must therefore be paid to positioning the centre of the circle. In an attempt to avoid bias in measurements, many authors have thought that 3D-CT methods could be more accurate, indeed Bois, et al. [17] comparing the accuracy of Glenoid Index [30], Ratio method [42] and PICO method [32] concluded that the PICO method is the best method albeit on 3D-CT. Magarelli, et al. [22] showed an agreement of 97% when calculating glenoid bone defects with the PICO method based on 2D and 3D-CT data. Therefore, to quantify glenoid bone defects any of these methods could be reproducible and accurate: the choice depends also on what kind of instrumentation is available. Above all, the crucial point is measurement accuracy: the glenoid has a width varying between 24-28 mm, so every 1.5-1.7 mm of missing bone correspond to 5% bone loss [39]. A defect of 6-8 mm results in bone loss of 20-25%, the critical value to opt for bone augmentation. For humeral head defects 3D-CT is the gold standard (Fig. 3), especially for their characterization and to quantify the bone loss. The location and size are the most important features also for Hill-Sachs lesions [38, 44, 45]. The typical location is on the postero-superior portion of the humeral head, in contrast to the bare area which is postero-laterally located [34]. The orientation is, however, also important, indeed in 2007 Yamamoto, et al. [38] introduced the concept of “glenoid track” to describe the contact zone between the glenoid and the humeral head during combined abduction and external rotation movements. There is a risk of engagement (Figs. 4a, b and c) with the glenoid rim if the head defect extends beyond the medial margin of said track. According to this theory, the depth and extent (width and length) are less important than location and orientation. If the long axis of the defect is parallel to the glenoid in abduction the external rotation engages its anterior rim, on the contrary if the long axis is diagonal in that arm position, it does not engage. In cases of intact glenoid the medial margin of the glenoid track is 18.4 ± 2.5 mm medial from footprint, and therefore a distance equivalent to 84 ± 14% of the glenoid width [38]. A recent study by Omori, et al. [46] has demonstrated that the width of the glenoid track at 60° of abduction was significantly greater than that with the arm at 90° or more. If a glenoid defect is concomitant, this could exacerbate the instability and increase the risk of a Hill-Sachs lesion to start engaging [47] and therefore reduce the area of the glenoid track, in contrast a larger Hill-Sachs could allow a smaller glenoid defect to cause engagement. Orientation was defined in relation to the deepest groove of the major axis of Hill-Sachs and the longitudinal axis of the humerus. The depth and length (and width though less importantly) are simply calculated both based on CT (Fig. 5) and MRI data, ideally 3D-CT with elimination of the scapula (en face humeral view). The length of the major axis was defined as the length of the Hill-Sachs lesion, while the length of the minor axis as its width (Fig. 6). On an axial view of the humeral head by CT or MRI (less accurate) a virtual circle that included the articular surface was drawn, and the depth of the lesion was defined as the longest length between the base of the Hill-Sachs lesion and the corresponding arc [20] (Fig. 7).

These two parameters are considered in the prediction of clinical severity and risk of recurrence or engagement. Rowe, et al. [40] have graded lesions as mild (< 0.5 cm deep / < 2 cm long) up to severe (> 1 cm deep / > 4 cm long). Flatow, et al. [48] have quantified humeral bone loss as a percentage of involvement of the entire head area on 2D-CT and classified lesions into three groups: < 20%, 20- 40% and > 40%. Hill-Sachs defects > 40% are clinically significant, and this is recognized as the critical value requiring bone graft, while 20% is the threshold for remplissage procedure, even if the lesion is not engaged [49-51]; the critical measurements for Hill-Sachs lesions are: a depth more than 16% of the head diameter, an area more than 20-25% of the humeral articular surface, a volume greater than 250 mm3 (calculated on 3D-CT) [49, 52, 53]. Area and volume may be more convenient to compare the size of Hill-Sachs lesions, and the best images are those obtained from en face 3D-CT.

Fig. (6). 3D-CT humeral en face view with scapula elimination: the most accurate method to quantify the humeral bone defect (the minor axis is the width measure and the major axis is the lenght) 49x56mm (150 x 150 DPI).

Fig. (7). Humeral bone defect depth measure on CT axial view with the circle method. 44x49mm (150 x 150 DPI).

Fig. (8). Altered signal intensity bone marrow oedema of the humeral head defect demonstrated on MRI. 61x58mm (150 x 150 DPI).

Di Giacomo, et al. [54] have recently proposed the concept of “on-track / off-track” lesions: if a Hill- Sachs lesion stays on the glenoid track (on-track lesion) the Bankart repair is recommended, if the lesion strays off-track the surgical procedure depends on the size of the glenoid defect. If it is smaller than 25%, remplissage combined with Bankart repair is recommended. But other authors have recently extended indications to perform remplissage even without substantial glenoid bone loss in large Hill-Sachs lesions and clinical instability [55].

The discovery of bipolar injury makes an accurate CT examination of both lesions even more imperative to avoid increased risk of recurrence: the glenoid and humeral head defects must be considered as two interrelated parts of the same problem.

MRI and MRA(conventional magnetic resonance and magnetic resonance arthrography).

Both MRI and MRA can be used for the identification and evaluation of bone defects (especially T2 weighted images), although non-displaced fractures can be difficult to detect after acute onset. However, it is in the quantification of defects that they show their greatest limit. Notwithstanding this, recent studies [43, 56-59] have demonstrated the validity and accuracy of both MRI and MRA in estimating glenoid bone defects compared with 2D and 3D-CT: MRA seems to be superior to MRI but its reliability is slightly lower than 3D-CT. Nonetheless, MRA is more sensitive and specific in presence of attritional bone loss compared to bony-Bankart investigations, and no study has reported MRI or MRA to be equal or superior to CT, especially 3D-CT, in the assessment of glenoid bone loss [45].

For humeral head defect detection the sensitivity of MRI is higher than that of CT and shows the greatest reproducibility [16]: this is due to the altered signal intensity of the bone marrow oedema demonstrated on both T1 and T2 weighted images, although fat-suppressed T2 weighted images are more sensitive (Figure 8). MRI shows high sensitivity (96.3%) and specificity (90.6%) for the presence of Hill-Sachs lesions (also chondral lesions), while MRA reaches a sensitivity of 100% even after surgical procedure. But the measurement of humeral defect size and the prediction of engagement or recurrent risk is an ambitious target for both MRI and MRA [60, 61]. Moreover, MRI can give false-positive results, misinterpreting erosions, small grooves or bare area for small Hill- Sachs lesions [34]; MRI importantly underestimates bone loss especially on the humeral side [62]. Furthermore, certain factors limit the large applicability of MRA in the quantification of bone loss: its high costs, invasiveness (relative), presence of ferromagnetic implants, possible allergic reactions and infections; on the other hand CT has the disadvantage of employing ionizing radiation.

Fig. (9). Grashey view: a true antero-posterior X-ray view in the scapular plane. 55x51mm (150 x 150 DPI).