Automatic orbital segmentation using deep learning-based 2D U-net and accuracy evaluation: A retrospective study.

The purpose of this study was to verify whether the accuracy of automatic segmentation (AS) of computed tomography (CT) images of fractured orbits using deep learning (DL) is sufficient for clinical application. In the surgery of orbital fractures, many methods have been reported to create a 3D anatomical model for use as a reference. However, because the orbit bone is thin and complex, creating a segmentation model for 3D printing is complicated and time-consuming. Here, the training of DL was performed using U-Net as the DL model, and the AS output was validated with Dice coefficients and average symmetry surface distance (ASSD). In addition, the AS output was 3D printed and evaluated for accuracy by four surgeons, each with over 15 years of clinical experience. One hundred twenty-five CT images were prepared, and manual orbital segmentation was performed in all cases. Ten orbital fracture cases were randomly selected as validation data, and the remaining 115 were set as training data. AS was successful in all cases, with good accuracy: Dice, 0.860 ± 0.033 (mean ± SD); ASSD, 0.713 ± 0.212 mm. In evaluating AS accuracy, the expert surgeons generally considered that it could be used for surgical support without further modification. The orbital AS algorithm developed using DL in this study is extremely accurate and can create 3D models rapidly at low cost, potentially enabling safer and more accurate surgeries.

Extension toward the Trapezius in a Transversely Oriented Latissimus Dorsi Flap for Breast Reconstruction.

A transverse paddle latissimus dorsi (LD) flap has the advantage that if the skin paddle is placed in the transverse bra line, the donor site scar is well hidden by underwear. With this transfer, medial back tissues are usually moved to the medial area of the reconstructed breast following 180 degree rotation. Because these tissues are thinner than the lateral thoracic area, the medial part of the reconstructed breast sometimes becomes flatter than expected. METHODS: To add bulk in the medial lower quadrant for giving an impression of an outward-expanding breast, we modified the LD flap by adding a part of the trapezius muscle. Seven patients underwent mastectomy and simultaneously received a modified LD flap. To hide the donor site scar beneath underwear, the skin paddle needed to be oriented transversely. The additional harvested tissues were tested for vascularity by fluorescence following intravascular injections of indocyanine green. If this was negative, the tissue was not used for breast reconstruction. Postoperatively, another surgeon judged whether this modification had contributed favorably to the reconstructed medial lower quadrant. RESULTS: Indocyanine green testing was positive in six cases. The shape of the lower medial quadrant was judged as good in five of the seven cases. Complications included an animation deformity of the LD muscle, donor site seroma, and donor site wound dehiscence. CONCLUSION: This transversely oriented LD flap with extension to the trapezius muscle placed at the bra-line is one option to add bulk to the medial lower quadrant of the reconstructed breast when an additional scar is not desired for cosmetic reasons.

Does an In-House Computer-Aided Design/Computer-Aided Manufacturing Approach Contribute to Accuracy and Time Shortening in Mandibular Reconstruction?

Mandibular reconstruction using computer-aided design and computer-assisted manufacturing (CAD/CAM) techniques has received recent attention. This technique has theoretical advantages, although this approach can be commercially used in the limited area of the world.The aim is to describe our experience using in-house CAD/CAM guides and the situations in which CAD/CAM may present benefit in the region where commercial guides are unavailable.The authors developed our In-house CAD/CAM approach for mandibular reconstructions with a free fibular flap. Patients were divided into 2 group; CAD/CAM and conventional groups. In the CAD/CAM group, reconstructions were planned virtually using CAD/CAM; these CAD/CAM guides were used in the surgery. In the conventional group, free-hand cutting and fitting of the fibular segments were performed as reconstructions. Later, the bone computed tomographic image was compared with the plan. The averaged deviations and the percentages of the points within 1 mm, 2 mm, and 3 mm deviations were recorded. Total and ischemic time were also recorded.Reconstruction points within 1 mm deviation were 59% of CAD/CAM group (n = 9) and 42% of conventional group (n = 10, P = 0.04), within 2 mm 82% and 69% (P = 0.03). Total time were 1012 and 911 minutes, while flap ischemic time were 147 and 175 minutes (P = 0.03), respectively.In-house CAD/CAM mandibular reconstruction also supported accuracy and shorter flap ischemic time. For a detailed accurate reconstruction, CAD/CAM showed superiority than conventional method. Use of the In-house CAD/CAM guides might be an option where commercial guides are not available.

Two Cases of the Vascular Territory of a Single-pedicled Deep Inferior Epigastric Perforator Flap with a Vertical Midline Abdominal Scar.

: The deep inferior epigastric perforator (DIEP) flap is becoming the gold standard for breast reconstruction using autologous tissue. If there are scars in the abdomen from previous surgery, it is necessary to judge the indication for using this flap carefully. Particularly in cases with vertical midline scars, the blood flow supply to the zone II can be compromised. Even when patients have a median abdominal scar, it has been reported that the blood flow can extend beyond the scar and reach several centimeters to about half of zone II. We performed breast reconstruction using DIEP flaps for 2 patients with vertical midline scars in the lower abdomen. Indocyanine green angiography was conducted intraoperatively to confirm the vascular territory with a single pedicle before cutting off the flap. One patient showed fluorescence contrast on the contralateral side across the midline scar. However, the fluorescence contrast was absent across the midline scar in the other patient. Based on this result, we investigated the possible vascular territory of a single pedicled DIEP flap in patients with vertical midline abdominal scars. We suggest that successful blood supply to zone II of a single-pedicled DIEP flap in a patient with a vertical midline abdominal scar is related to the location of the perforator and the property of the tissue in the midline near the perforator. However, because it is difficult to predict the vascular territory of a single pedicle before surgery, intraoperative evaluation using such techniques such as indocyanine green fluorescence imaging is important.

Designing CAD/CAM Surgical Guides for Maxillary Reconstruction Using an In-house Approach.

Computer-aided design/computer-assisted manufacturing (CAD/CAM) is now being evaluated as a preparative technique for maxillofacial surgery. Because this technique is expensive and available in only limited areas of the world, we developed a novel CAD/CAM surgical guide using an in-house approach. By using the CAD software, the maxillary resection area and cutting planes and the fibular cutting planes and angles are determined. Once the resection area is decided, the necessary faces are extracted using a Boolean modifier. These superficial faces are united to fit the surface of the bones and thickened to stabilize the solids. Not only the cutting guides for the fibula and maxilla but also the location arrangement of the transferred bone segments is defined by thickening the superficial faces. The CAD design is recorded as .stl files and three-dimensionally (3-D) printed as actual surgical guides. To check the accuracy of the guides, model surgery using 3-D-printed facial and fibular models is performed. These methods may be used to assist surgeons where commercial guides are not available.

Using an In-House Approach to Computer-Assisted Design and Computer-Aided Manufacturing Reconstruction of the Maxilla.

PURPOSE: Computer-assisted design (CAD) and computer-aided manufacturing (CAM) techniques are in widespread use for maxillofacial reconstruction. However, CAD/CAM surgical guides are commercially available only in limited areas. To use this technology in areas where these commercial guides are not available, the authors developed a CAD/CAM technique in which all processes are performed by the surgeon (in-house approach). The authors describe their experience and the characteristics of their in-house CAD/CAM reconstruction of the maxilla. PATIENTS AND METHODS: This was a retrospective study of maxillary reconstruction with a free osteocutaneous flap. Free CAD software was used for virtual surgery and to design the cutting guides (maxilla and fibula), which were printed by a 3-dimensional printer. After the model surgery and pre-bending of the titanium plates, the actual reconstructions were performed. The authors compared the clinical information, preoperative plan, and postoperative reconstruction data. The reconstruction was judged as accurate if more than 80% of the reconstructed points were within a deviation of 2 mm. RESULTS: Although on-site adjustment was necessary in particular cases, all 4 reconstructions were judged as accurate. In total, 3 days were needed before the surgery for planning, printing, and pre-bending of plates. The average ischemic time was 134 minutes (flap suturing and bone fixation, 70 minutes; vascular anastomoses, 64 minutes). The mean deviation after reconstruction was 0.44 mm (standard deviation, 0.97). The deviations were 67.8% for 1 mm, 93.8% for 2 mm, and 98.6% for 3 mm. The disadvantages of the regular use of CAD/CAM reconstruction are the intraoperative changes in defect size and local tissue scarring. CONCLUSION: Good accuracy was obtained for CAD/CAM-guided reconstructions based on an in-house approach. The theoretical advantage of computer simulation contributes to the accuracy. An in-house approach could be an option for maxillary reconstruction.

Dual Vascular Free Anterolateral Thigh Flap.

BACKGROUND: The optimum number of microvascular anastomoses for safe free tissue transfer is controversial. Although the case for 2 venous anastomoses versus 1 anastomosis has been argued, the use of an additional arterial anastomosis has not been examined in detail. METHODS: Twelve patients who underwent 2 arterial anastomoses for a free flap transfer were identified retrospectively from the medical records of patients undergoing reconstruction for head and neck cancer. The free flaps were limited to anterolateral thigh (ALT) flaps. RESULTS: All flaps survived. Complications included venous thrombosis (n = 1), reexploration (n = 1), and leakage (n = 3). The vascular patterns of dual-arterialized ALT flaps were classified into 3 groups. Types 1 and 2 were ALT flaps that had 2 vascular sources from the descending and lateral branches of the lateral circumflex femoral artery. The number of accompanying veins differed between type 1 (3 veins) and type 2 (2 veins). Type 3 differed from a conventional ALT flap nourished by the descending branch of the lateral circumflex femoral artery (1 vein) by the addition of anastomosis of an artery branching from the descending branch to the vastus medialis muscle. The total operation times for these 3 types of ALT were similar. CONCLUSIONS: An additional arterial anastomosis to the free cutaneous flap did not cause any congestion or disturb the balance between inflow and outflow. If the surgeon considers that the first arterial anastomosis is unreliable, an additional anastomosis might be an option in ALT transfer.

Secondary Maxillary and Orbital Floor Reconstruction With a Free Scapular Flap Using Cutting and Fixation Guides Created by Computer-Aided Design/Computer-Aided Manufacturing.

Computer-aided design/computer-aided manufacturing (CAD/CAM) guides are now widely used in maxillofacial reconstruction. However, there are few reports of CAD/CAM guides being used for scapular flaps. The authors performed the secondary maxillary and orbital floor reconstruction using a free latissimus dorsi muscle, cutaneous tissue, and scapular flap designed using CAD/CAM techniques in a 72-year-old man who had undergone partial maxillectomy four years previously. The patient had diplopia, the vertical dystopia of eye position, and a large oral-nasal-cutaneous fistula. After the operation, the authors confirmed that the deviation between the postoperative and preoperative planning three-dimensional images was less than 2 mm. Because scapular guides require 3 cutting surfaces, the shape of the scapular guide is more complex than that of a conventional fibular guide. In orbital floor reconstruction, the use of a CAM technique such as that used to manufacture the authors' fixation guide is as necessary for accurate, safe, and easy reconstruction as is preoperative CAD planning. The production of a fixation guide as well as a cutting guide is particularly useful because it is difficult to determine the angle for reconstructing the orbital floor by freehand techniques. In this case, the orbital floor was reconstructed based on a mirror image of the healthy side to avoid overcompression of the orbital tissue. Although the patient's vertical dystopia of eye position was improved, diplopia was not improved because, for greater safety, the authors did not plan overcorrection of the orbital volume.

Intraoperative Change in Defect Size during Maxillary Reconstruction Using Surgical Guides Created by CAD/CAM.

Surgical osteotomy guides created by computer-aided design/computer-aided manufacturing (CAD/CAM) have been developed and are now widely used in maxillofacial reconstruction. However, there are no standard procedures for dealing with an intraoperative change in defect size. We report on a case in which we used our CAD/CAM guides to deal with an intraoperative change in defect size in a maxillary reconstruction. We planned the maxillary reconstruction using a free fibula flap because of left maxillary sinus cancer in a 73-year-old man. In Japan, we cannot use commercially supplied CAD/CAM guides because these have not been approved by the government. We created novel CAD/CAM guides by using free software and a low-cost 3D printer. We performed model surgery to check the accuracy of the design and to prebend the titanium plates before the operation. The actual defect in the maxilla was found to be smaller than that used in preoperative planning. It was therefore necessary to rearrange the fibular segments and to rebend the plates. Comparison between the preoperative and postoperative 3D images showed that the deviation was 2-4 mm. In case that the CAD/CAM guides become inapplicable because of an intraoperative change in defect size, rearranging both the ends of set-up fibular segments and rebending the plates in situ allows us to deal with the situation. However, because extra time is needed to rearrange and rebend, the total operation and flap ischemic times are not shortened.