The accuracy of navigated versus freehand curettage in bone tumors: a cadaveric model study

Abstract

Purpose: Navigation has been suggested to guide complex benign bone tumor curettage procedures, but the contribution of navigation to the accuracy of curettage has never been quantified. We explored the accuracy of navigated curettage in a cadaveric observational pilot study, comparing navigated to freehand curettage, performed independently by an expert and a novice user.

Methods: The expert performed curettage on 20 cadaveric bones prepared with a paraffin wax mixture tumor, 10 freehand and 10 navigated. We re-used 12 bones for the novice experiments, 6 freehand and 6 navigated. Tumor and curettage cavity volumes were segmented on pre- and post-cone-beam CT scans. Accuracy was quantified using the Dice Similarity Coefficient (DSC), and with remaining tumor volume, bone curettage volume, maximal remaining width and procedure times compared between navigation and freehand groups for both users.

Results: There were little differences in curettage accuracy between a navigated (DSC 0.59[0.17]) and freehand (DSC 0.64[0.10]) approach for an expert user, but there were for a novice user with DSC 0.67(0.14) and 0.83(0.06), respectively. All navigated and freehand procedures had some amount of remaining tumor, generally located in a few isolated spots with means of 2.2(2.6) cm³ (mean 20% of the tumor volume) and 1.5(1.4) cm³ (18%), respectively, for the expert and more diffusely spaced with means of 5.1(2.8) cm³ (33%) and 3.0(2.2) cm³ (17%), respectively, for the novice.

Conclusions: In an explorative study on 20 cadaveric bone tumor models, navigated curettage in its current setup was not more accurate than freehand curettage. The amount of remaining tumor, however, confirms that curettage could be further improved. The novice user was less accurate using navigation than freehand, which could be explained by the learning curve. Furthermore, the expert used a different surgical approach than the novice, focusing more on removing the entire tumor than sparing surrounding bone.

Keywords: Bone; Computer-assisted surgery; Curettage; Navigation; Orthopedic; Tumor.

Fig. 1

The cadaveric bone tumor model, with a being the situation just after filling the two cavities for bone 3. The paraffin wax mixture has started to solidify around the edges, but is still fluid in the center; b some examples of finished cadaveric bone tumor models, with bone 3 in the upper left corner

Fig. 2

The experimental setup during navigated curettage of a distal femur. The surgeon uses the tracked pointer to assess the tumor shape and location

Fig. 3

Cone-beam CT scans of the cadaveric bone tumor model; a model with good contrast between tumor and bone on the pre-CBCT (bone 3); b model with poor contrast between tumor and bone on the pre-CBCT; c the removed tissue for bone 3 on the post-CBCT, with tumor remaining at the proximal and distal ends; d the fused pre-CBCT (dark) and post-CBCT (light) of bone 3 with the segmentations of the tumor (blue) and curettage cavity (green), highlighting the remaining tumor parts

Fig. 4

Graphs showing key outcomes for each individual bone in the navigated and freehand groups and the expert and novice experiments. The bone ID represents the code for each bone which was kept the same for both user experiments, it does not indicate the treatment order. Bone 15 was excluded from the expert experiments as the two bone halves separated during curettage, but it could be re-used for the novice experiments together with eleven other bones (bone 2, 3, 6, 7, 10, 11, 13, 15, 16, 17, 18, 19); a The accuracy using the Dice Similarity Coefficient (DSC^Ac); b the absolute remaining tumor volume and c the relative remaining tumor volume as a percentage of the initial tumor volume

Fig. 5

Grouped scatterplots illustrating the effect of navigation on key outcome values for the expert and novice user, with means indicated by the horizontal lines. a The accuracy using the Dice Similarity Coefficient (DSC^Ac); b the absolute remaining tumor volume; and c the relative remaining tumor volume as a percentage of the initial tumor volume

Fig. 6

Visualizations of the remaining tumor in various bones; a Is a 3D model showing the remaining tumor in a few isolated spots in the expert bone 3. The curettage cavity is depicted in gray, and the remaining tumor is colorized, with green being a small width and dark blue being the maximal remaining tumor width of in this case 5.1 mm; b Illustrates the more diffusely spaced remaining tumor in the novice bone 11. The color range is the same as in (a)