Introduction
Odontogenic cysts are the most frequent lesions appearing in the jaws. They are defined as cavities filled with liquid, semiliquid, or gaseous content with odontogenic epithelial lining and connective tissue on the outside. They originate from the epithelial component of the odontogenic apparatus or its remnants that lie entrapped within the bone or in the peripheral gingival tissues [
1].
Most cysts of the jaws are discovered incidentally on panoramic radiographs or they destroy surrounding structures and cause problems such as loosening of teeth or facial deformity. Panoramic radiograph is often routinely used as a primary diagnostic tool for the detection of cystic lesions, particularly in follow-up to assess neo-ossification. In more complex maxillofacial surgical cases requiring 3D information of the region of interest, CBCT offers advantages over conventional 2D imaging modalities, such as a detailed representation of cysts in bone tissue and the involvement of surrounding structures, such as tooth roots and nerves. For this reason, they allow surgeons to accurately plan surgical management [
2].
According to cyst size, jaw cysts can be classified into small, median, and large mandibular cysts, which often invade teeth and can seriously affect the quality of life. Cysts tend to enlarge and grow, leading to resorption of bone tissue. Depending on the degree of resorption, cysts may cause severe damage such as bone fractures. Treatment planning for cysts depends on cyst location, size, extent of tissue damage, availability of surgical access, patient’s age, proximity of the cyst to vital structures, and significance of the affected teeth in terms of eruption. Marsupialization or decompression is the first consideration if the lesion invades adjacent structures or if primary enucleation could cause pathological fractures or neurological damage [
3,
4].
Small cysts can generally achieve satisfactory results after root canal treatment, while curettage is an effective and radical treatment for median and large cysts. However, large cysts (> 4 cm in diameter) lead to a large area of involvement and more critical anatomical damage and are more likely to cause problems, such as bone destruction, maxillofacial deformity, and can affect occlusal function. Therefore, conventional cyst curettage is not effective enough [
5].
Non-invasive determination of the volume of the lower jaw cysts is a helpful additional process in the preoperative diagnosis. In this way, a geometrical approximated volume can be calculated. Linear measurements on CBCT images are possible in all three planes and directions and are employed in routine practice. Volumetric studies on bone regeneration of cystic cavities were carried out using CT scanning and measuring the three maximum diameters of the cavity. Nonetheless, it is still an approximated volume, not a real volumetric measurement. From these diameters, approximate volumes were calculated using the cubic and ellipsoid formulas. The data suggest that maximum tumor diameter-based size characterization, especially the cuboid formula and the maximum diameter alone, should not be recommended [
6].
Image segmentation is used to analyze and process 2D or 3D images to achieve extraction, 3D reconstruction, and 3D visualization of anatomical structures or anomalies such as tumors or cysts. Volumetric analysis requires segmentation of an object, such as a tooth, from its surrounding structures. With the help of image segmentation, the physician is provided with a tool to determine the volume of a jaw lesion, and, in addition to that, the anatomical extent can be clearly defined and used for surgical planning. The volume of affectation of caries is determined with the k-mean grouping method and the threshold method, the latter being the most recommended [
7].
Volumetric analysis within the field of dental–maxillofacial radiology can be utilized for assessing volumetric estimates of different bone injuries counting; periapical abscesses, cysts, and tumors. Identifying the volume of a lesion is vital, particularly in comparing the measurements with the follow-up radiographs. The error of volumetric measurement in CBCT reconstruction may have an important clinical impact. The inaccuracy of volumetric measurement can influence superimposition and comparison before and after surgery [
8,
9].
A growing number of software programs to manage and analyze Digital Imaging Communications in Medicine (DICOM) files are available in the market every year. Many of these have incorporated tools for segmentation and volumetric analysis. Several software packages already provide clinicians with a dedicated tool for assessing the volumes of regions of interest in cubic millimeters. Several previous studies have focused on the accuracy of volume measurements of teeth from CBCT data [
10,
11]. Therefore, is the formula-based volume of cystic lesions comparable to values to the volumetric analysis values? Is the volumetric analysis of cystic jaw lesions affected by the change in software?
Thus, the aim of this study was to assess the accuracy of formula-based volume measurements and 3D volume analysis with different software packages in the calculation of preoperative cystic jaw lesions’ volume. The secondary aim was to assess the reliability and the accuracy of 3 imaging software programs for measuring the cystic jaw lesions' volume in CBCT images.
Discussion
Researchers have utilized linear measurements of the CBCT scans for determining the volumes of periapical defects. Linear measurements provide limited information and values in one plane in comparison to volumetric measurements. As the periapical lesions may have irregular or 3D shapes, the algorithm for calculating the volume of the sphere cannot be applied. Linear accuracy of the measurements done by this method is inadequate to translate into a clinical setting [
15‐
17].
In this study, the DICOM data from CBCT scans were segmented using OnDemand, MIMICS, and InVesalius software. OnDemand is the software on a local workstation. MIMICS was chosen because of its widespread use in Biomedical engineering. MIMICS software provides semi-automatic segmentation and interpolation between slices and detects the margins to save time. InVesalius software was included according to its easy availability for everybody on a free open-source basis.
In this study, there was a statistically significant difference between volumetric measurements by equation and different modalities (
P value < 0.001). All software showed statistically significantly higher volumetric measurements than measurements by equation. This is consistent with Kauke et al. [
5] who investigated the agreement and overlap between image segmentation and formula-based volume approximation. The ellipsoid formula yielded volume approximations that were in mean 10.1% lower when compared to segmentation-based volume approximations using ITK-snap software. They inferred that formula-based volume approximation is error-prone and not precise when compared to image segmentation as odontogenic jaw lesions rarely grow in the perfect shape of a cuboid or ellipsoid, both formulas are naturally error-prone. In particular, this applies to infiltrative odontogenic neoplasms, capable of arbitrary three-dimensional infiltration with budding and thus irregular three-dimensional configuration.
This can be explained by Lizio et al. [
18] who stated that calculating the area of a lesion as a regular ellipsoid is an approximation that does not take into consideration the frequent morphological irregularities of these lesions, especially keratocyst, and the presence of collateral cavities and scalloped contours. Its shortcomings include inaccurate discrimination of cyst border and the inability to assess the cyst’s relation with the surrounding vital structures. Furthermore, the ellipsoid formula depends on bi-dimensional evaluation and measurements of the largest dimensions of the diameter and depth of the lesion. Observer performance, selection of reference points, mouse sensitivity, and software capabilities are all important factors in the measurement of lesion dimensions. In this study, experienced and calibrated oral radiologists equally familiar with the software used acted as operators [
18,
19].
On the other hand, Dejaco et al. [
12] found that an ellipsoid formula using the largest diameter of a lesion in all three planes provided a reasonable approximation of head and neck tumor volumes when compared to manual slice-by-slice segmentation. However, the use of such a mathematical formula can be cumbersome in multi-locular lesions. Also, Sacher et al. [
20] who used OsiriX software, concluded that using the formula is easy to use and allows for an accurate and precise prediction of the amount of time needed for bone regeneration after both cystostomy and cystectomy. This means that the formula can be used for comparable conditions.
Concerning the operator-dependent error, manual and semi-automatic segmentation showed very good inter-operator reliability according to the ICC values and the small volumetric differences found between the three recordings. The highest inter-examiner agreement for volume measurement was found with manual segmentation using InVesalius software (Cronbach’s alpha = 0.992). This was followed by semi-automated segmentation using MIMICS software (Cronbach’s alpha = 0.989) and OnDemand software (Cronbach’s alpha = 0.963).
In this respect, the segmentation process delegated most of this task to the software algorithm, which reduced the magnitude of the observer-related error. For the same reason, semi-automatic segmentation almost voided the difference between the readings performed by three observers with different level of expertise in 3D imaging. In addition, threshold adjustment is solely dependent on the operator; thus, checking the integrity of the segmented object on 3 spatial planes is crucial. Although the aforementioned threshold adjustments could have created differences between the operators, the results of this study still represent high inter-operator. It was concluded that the selection of threshold sensitivity values was not reliable [
21].
Also, Weissheimer et al. [
10] compared the precision and the accuracy of 6 imaging software programs (Mimics, Dolphin3D, Ondemand3D and ITK-Snap, and InVivo Dental). The method repeatability for the patients' oropharynx measurements was high (ICC 0.0.94) for 6 imaging software programs. In addition, Chen et al. [
22] assessed the reliability and accuracy of three different commercially available software packages (Amira, 3Diagnosys, and OnDemand3D). The intra- and inter-observer reliability of the measurements using all three software packages were excellent (ICC ≥ 0.75). All three software packages generally underestimated the upper airway volume. In Abdelhamid et al. [
23,
24] study, the inter-observer reliability was high for OnDemand and InVesalius programs, which indicated minimal subjective variance for well-trained practitioners. Also, ElShebiny et al. [
25] reported high reliability was observed between four tested software packages including OnDemand3D for intra-operator and inter-operator values.
In this study, MIMICS software measurements showed higher mean volumes (3244.3 mm
3) than OnDemand (3209.2 mm
3). This is consistent with Weissheimer et al. [
10] also assessed segmentation with interactive thresholding using 6 software including MIMICS and OnDemand. The volume measurements with the 6 imaging software were statistically different (
P = 0.006). The descriptive statistical analysis showed higher oropharynx mean volumes for MIMICS and lower mean volumes for Ondemand3D. There were no statistically significant differences (
P > 0.05) among ITK-Snap, Mimics, OsiriX, Dolphin3D, and Ondemand3D.
These results are in good agreement with a previous study by El H & Palomo. [
26] who reported that OnDemand3D software sometimes fails to depict certain parts of the upper airway, which subsequently leads to an underestimation of the airway volume. This phenomenon could originate from the CBCT image acquisition process and/or the subsequent image segmentation by means of thresholding. During CBCT image acquisition, anatomical structures are discriminated based on their radiographic density. However, voxels residing on tissue boundaries can contain more than one tissue type. This phenomenon is known as the partial volume effect. The result of the partial volume effect is that voxels are erroneously allocated to “soft tissue” instead of “air” and hence “upper airway” during the image segmentation process [
24,
27].
This may be explicated by Lo Giudice et al. [
21] who inferred that software based on a threshold-based segmentation algorithm (MIMICS, OnDemand, and InVesalius) could cause an under/overestimation of boundaries since the segmentation procedure still relies on the operator visual discrimination of the bony structure and definition of threshold-level. Consequently, if an accurate definition of an object’s boundaries is required, highly skilled clinicians can perform manual refinement.
In this study, pair-wise comparisons revealed that there was no statistically significant difference between Mimics, OnDemand, and InVesalius software. Abdelhamid et al. [
23] also showed that OnDemand and InVesalius software had comparable volumetric computation in the presurgical volumetric analysis in secondary alveolar cleft bone grafting, but InVesalius was relatively faster than On-Demand 3D. To select the best option, it is necessary to analyze not only the time spent in the process, knowing that time is important, but accuracy and robustness of the software [
28]. In addition, Ghoneim and Gad [
29] tested the difference between measures carried out by MIMICS and AutoCAD software for the post-marsupialization of cystic lesions, and the results were non-significant (
P > 0.05).
In conclusion, volumetric measurements are influenced by the software's imaging processing methods and segmentation techniques, and these differ between the different software. This current study shows that the 3D computer-aided assessment of cyst volumes provides information that is accurate enough to be used for preoperative planning. In addition, further investigations of volumetric analysis of cystic jaw lesions will be needed to correlate with different shapes of cystic lesions. Also, further investigation of the effect of the preoperative size of cystic jaw lesions and proper grouping of the sizes will be valuable.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.