based on an association with conventional radiography features and clinical classification
Yoshiko Ariji, DDS, PhD,aMasahiro Izumi, DDS, PhD,aMasakazu Gotoh, DDS,b Munetaka Naitoh, DDS, PhD,cMugio Katoh, DDS, PhD,cYuichiro Kuroiwa, DDS,d Naofumi Obayashi, DDS,eKenichi Kurita, DDS, PhD,fKazuo Shimozato, DDS, PhD,g and Eiichiro Ariji, DDS, PhD,hNagoya, Japan
AICHI-GAKUIN UNIVERSITY SCHOOL OF DENTISTRY AND MUGI DENTAL CLINIC
Objectives. The aims of this study were (1) to assess the diagnostic power of magnetic resonance imaging (MRI) for mandibular osteomyelitis through comparison with conventional techniques and (2) to establish practical MRI diagnostic criteria in relation to treatment and clinical outcome.
Study design. In 55 subjects, clinically suspected as mandibular osteomyelitis, signal intensities (SI) were evaluated on T1-weighted/short T1 inversion recovery (STIR) images.
Results. Forty-seven subjects were definitively diagnosed as having osteomyelitis by pathology studies or clinical course. For the acute or subacute stage, positively associated appearances were low SI on T1-weighted image and extensive high or focal high SI on the STIR image. For chronic stage, appearances of low SI on both T1-weighted and STIR images should be added to those for the acute or subacute stage. These findings support the at-present accepted imaging diagnostic criteria based on bony changes for detection of osteomyelitis.
Conclusion. This study confirms that T1-weighted/STIR images are useful for the detection of mandibular osteomyelitis. (Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2008;105:503-11)
Osteomyelitis is defined as an inflammation of the cortical and cancellous bone and is commonly caused by bacterial invasion from contiguous foci.1,2 Radio- logical examinations are required for diagnosis of man- dibular osteomyelitis and for evaluation of the response to treatment. On conventional radiographs, osteomyeli- tis appears as an osteolytic or osteosclerotic lesion with various amounts of periosteal reaction. However, their
utility is limited in the acute stage because of low sensitivity.3,4 Bone scintigraphy can depict increased uptake even in the acute stage, whereas it shows low specificity in diagnosing as osteomyelitis and does not accurately identify the location of lesions.4-6Computed tomography (CT) is known to be the most effective tool for the evaluation of bony changes, such as cortical bone lesion, sequestra, and subperiosteal bone deposi- tion, whereas it can hardly depict bone marrow changes in the acute stage or evaluate the early response to the treatment.3,4,6-8
Magnetic resonance imaging (MRI) can well dem- onstrate the bone marrow changes caused by edema or inflammatory tissue due to increase of water content, which often replaces the normal fatty marrow in the acute stage.5This change of the bone marrow is shown as a low SI area on the T1-weighted image and a marked high SI area on T2-weighted or short TI inver- sion recovery (STIR) images.3,4,6,9,10 In the chronic stage, sequestration as a result of necrosis is observed, and its MRI appearance is somewhat different from that in the acute stage. It is characterized by a low SI area surrounded by high SI rim on both T1-weighted images and T2-weighted or STIR images.4Although MRI ex- amination has the ability of early detection of osteo- myelitis, almost like bone scintigraphy,4,5,11,12the role of MRI examination in the clinical workup of mandib- ular osteomyelitis has not still been established. There
aAssistant Professor, Department of Oral and Maxillofacial Radiol- ogy, Aichi-Gakuin University School of Dentistry.
bInstructor, Department of Oral and Maxillofacial Radiology, Aichi- Gakuin University School of Dentistry.
cFormerly Assistant Professor, Department of Oral and Maxillofacial Surgery I, Aichi-Gakuin University School of Dentistry; Currently Director, Mugi Dental Clinic.
dInstructor, Department of Oral and Maxillofacial Surgery I, Aichi- Gakuin University School of Dentistry.
eInstructor, Department of Oral and Maxillofacial Surgery II, Aichi- Gakuin University School of Dentistry.
fProfessor and Chairman, Department of Oral and Maxillofacial Surgery I, Aichi-Gakuin University School of Dentistry.
gProfessor and Chairman, Department of Oral and Maxillofacial Surgery II, Aichi-Gakuin University School of Dentistry.
hProfessor and Chairman, Department of Oral and Maxillofacial Radiology, Aichi-Gakuin University School of Dentistry.
Received for publication Jan 21, 2007; returned for revision Mar 16, 2007; accepted for publication Apr 25, 2007.
1079-2104/$ - see front matter
© 2008 Mosby, Inc. All rights reserved.
doi:10.1016/j.tripleo.2007.04.029
503
is a relative dearth of reports regarding MRI diagnostic criteria applicable to each clinical stage of osteomyeli- tis.2 To establish practical criteria, the stage (acute, subacute, or chronic) should be taken into account.
The purposes of this study were (1) to analyze the relationship between the imaging features of MRI and conventional techniques and (2) to establish practical MRI diagnostic criteria in relation to the treatment outcomes.
METHODS Subjects
Fifty-five subjects who were clinically suspected as having bacterial mandibular osteomyelitis and under- went MRI examination were enrolled in this study.
They were 24 males and 31 females, ranging from10 to 97 years (median: 43 years) of age. All subjects were selected retrospectively from the files of our department and were imaged between August 1996 and July 2006.
Criteria for inclusion into this study were recurrent and intolerable pain or swelling at the mandibular region and radiographic evidence of osteolytic and osteoscle- rotic lesion with variable amounts of periosteal reaction or sequestration. A widespread sclerosis extending in- feriorly beyond the mandibular canal and posteriorly to the ramus was also included. Twenty-seven subjects had recurrent swelling and intolerable pain in the man- dibular region. In 26 subjects, paralysis was accompa- nied at the lower lip, in addition to recurrent swelling and intolerable pain. In the remaining 2 subjects, the main symptoms were slight swelling and prolonged dull pain, but their radiographs showed a widespread sclerosis extending inferiorly beyond the mandibular canal and posteriorly to the ramus. Thirteen subjects were accompanied with trismus. Subjects with osteora- dionecrosis or anamnesis of surgical procedure for os- teomyelitis were not included in the study.
The period between the onset of symptoms and MRI examination ranged from 4 days to 5 years (median, 2 months). Based on this period, all subjects were clas- sified into 2 groups: the acute or subacute group (23 subjects) with the period less than 1 month, and the chronic group (32 subjects) with the period more than 1 month.
Examinations
Magnetic resonance imaging examinations were per- formed with a 1.0-T Magnex-100XP (Shimadzu, Kyoto, Japan) and a head coil. Standard MRI sequences were performed: spin-echo T1-weighted images (450- 500/11-18 [repetition time ms/echo time ms]) and STIR images (2800-3200/22/90-110 [repetition time ms/echo time ms/inversion time ms]). The section thickness was 5.0 mm with an intersection gap of 1.0 mm. The ac-
quisition matrix was 256 ⫻ 256. Axial and coronal images were obtained.
Prior to MRI examination, conventional radiographs (panoramic radiograph, lateral oblique mandible and posteroanterior view) were obtained for all subjects.
Computed tomographic examination was added in 41 subjects. All subjects, excepting those with the definite bone changes and sequestrum, underwent CT examina- tions. Computed tomographic images were obtained with a Somatom ART (Siemens AG, Erlangen, Ger- many) or a HiSpeedNX/Ipro (GE Medical Systems, Tokyo, Japan). Scans were performed with a slice thickness of 2 or 3 mm, and the scan plane was parallel to the occlusal plane or inferior margin of the mandible.
Fig. 1. Patterns based on the change of the signal intensity (SI) of the bone marrow on T1-weighted images. A, Low. The socket of the extracted tooth showing middle SI, extensive low SI area of the left mandible, and swelling of the adjacent buccal soft tissue. B, No change (arrow).
Evaluation of imaging features
According to previous reports,3,4,6,9,10 T1-weighted images were classified into 2 patterns: low and no change (Fig. 1). The pathological changes of the bone marrow were evaluated by comparison with the SI of the contralateral side. Low was defined as a decline in the SI of the bone marrow, and no change as a mild change or no change in the SI of the bone marrow.
Short T1 inversion recovery images were classified into 4 patterns: extensive high, focal high, low, and no change (Fig. 2). Extensive high was defined as a strong and widespread increase in the SI of the bone marrow.
Focal high was a limited area increase in the SI. Low was defined as a decline in the SI of the bone marrow, and no change as a mild change or no change in the SI of the bone marrow.
Bony changes on conventional images were evaluated with reference to the CT appearance (Fig. 3).3,4,6-8Lytic pattern was defined as having a feature that osteolysis was dominant. Mixed pattern was a feature that osteosclerosis was combined with an equal or lesser amount of osteol- ysis. Sclerotic pattern was a feature in which osteoscle- rosis was dominant. This included the cases in which the sclerotic change spread extensively under the mandibular canal or through the ramus. Sequestrum pattern was a feature in which sequestration was seen with or without other bony changes. The subjects without any such changes were recorded as no change.
These evaluations were performed by 2 radiologists who had sufficient experience in imaging diagnosis.
The final determination was reached by consensus after discussion if the evaluations initially differed between the 2 observers.
Fig. 2. Patterns based on the change of SI of the bone marrow on short T1 inversion recovery (STIR) images. A, Extensive high.
Shows erosion of the cortical bone (arrow) and the adjacent soft tissue swelling with high SI. B, Focal high (arrow). C, Low (arrow). D, No change (arrow).
Determination of the final diagnosis
The definitive diagnosis was confirmed by the patho- logical results or clinical course. All subjects were followed up for more than 6 months after the end of treatment. For example, subjects who achieved relief from symptoms after the surgical procedure for osteo- myelitis, such as decortication and fenestration, were diagnosed as osteomyelitis.
RESULTS
Treatment and the final diagnosis
Forty-seven subjects were finally diagnosed as os- teomyelitis of the mandible. Of them, 17 subjects un- derwent radical surgical procedure such as decortica- tion or fenestration of the buccal cortical bone of the mandible, accompanied by curettage of the affected areas, intra-arterial use of antibiotics, or hyperbaric oxygen therapy. For 23 subjects, minor surgical proce- dure including curettage, sequestrectomy, extraction of
the causative tooth, and incision of abscess, was per- formed. Seven subjects were treated only with internal use of antibiotics. Fifteen subjects who underwent sur- gical treatment were confirmed as osteomyelitis by pathological specimens. In 41 (87%) of these 47 sub- jects, symptoms disappeared or markedly reduced after the treatment and did not recur within the observation period. In 5 subjects, symptoms disappeared once but recurred. The remaining 1 subject was not cured by the treatment, but the pathological specimen obtained dur- ing surgical procedure verified the diagnosis of osteo- myelitis.
In contrast, 8 subjects could not be diagnosed as osteomyelitis. Of these, 3 subjects were prescribed car- bamazepine or received stellate ganglion block, leading to the symptoms temporarily disappearing or being reduced. In 3 subjects, the symptoms disappeared spon- taneously within 2 months; their symptoms might have been due to localized periodontitis or the healing pro- Fig. 3. Patterns of conventional imaging features. A, Lytic. B, Mixed. C, Sclerotic. D, Sequestrum. Ranges with changes in bone marrow are indicated with arrows.
Fig. 4. 47-year-old man. The patient had dull pain in the cheek and the buccal gingiva of the right mandibular molar region for 2 weeks. He complained of paralysis in the lower lip from the day before admission. A and B, Panoramic tomography and computed tomographic image did not show any change of the cancellous and cortical bone in the right mandibular molar region.
On the next day, MRI examination was done. C, The T1-weighted image showed a low SI area in the molar to ramus region of the right mandible (arrow). D, The STIR image showed an extremely high SI area in the same region (arrow).
Table I. Relationship between MRI features and bony appearance in 23 patients in the acute or subacute group
MRI features Bony appearances on conventional radiography and CT
T1-weighted image
T2-weighted or
STIR image Lytic Mixed Sclerotic Sequestration No change
Low Extensive high 2 4 3 1 6
Focal high 1 4
Low No change No change Extensive high
Focal high 2*
Low No change
Total 2 5 7 1 8
MRI, magnetic resonance imaging; CT, computed tomography; STIR, short TI inversion recovery.
*These 2 patients could not be diagnosed as osteomyelitis.
cess after tooth extraction. The remaining 2 subjects probably had mental problems, such as cancer phobia.
In all these subjects, there were no definitive appear- ances of osteomyelitis on conventional radiographs.
Relationship between MRI and conventional imaging features
The relationships between MRI and conventional imaging features in 23 subjects of the acute or sub- acute group are shown in Table I. Twenty-one sub- jects had low SI on T1-weighted imaging. Of these, 16 subjects had extensive high SI and 5 had focal high SI on T2-weighted or STIR imaging. These 21 subjects were diagnosed as having osteomyelitis based on pathological specimens or clinical outcome.
In 6 of these subjects, MRI appeared to be an effec- tive tool for the diagnosis because there were no apparent bony changes on conventional radiography and CT (Fig. 4). Two subjects, who could not be diagnosed definitively as having osteomyelitis, showed no change on T1-weighted imaging and focal high SI on STIR imaging. No bony changes were observed in these 2 subjects.
In the chronic osteomyelitis group, 26 of 32 subjects had low SI on T1-weighted imaging (Table II). Of these, 19 subjects had an extensive high, and 6 subjects had focal high SI on STIR imaging (Fig. 5). One subject showed low SI both on T1-weighted and STIR imaging (Fig. 6). This appearance could not be ob- served in the acute or subacute stage and might well have been caused by widespread sclerotic change, which was characteristic in the chronic osteomyelitis.
These 26 subjects could be diagnosed definitively as having osteomyelitis. Four of them showed significant findings for the diagnosis solely on MRI. The remain- ing 6 subjects, who could not be diagnosed as osteo- myelitis, showed no change on T1-weighted imaging
and in CT bony appearance. Two of them had no findings in any of the images evaluated.
Proposal of criteria for MRI
Based on the aforementioned results, we propose the following diagnostic criteria, using MRI, for osteomy- elitis: for the acute or subacute stage, the positive appearance of osteomyelitis is low SI on T1-weighted imaging, together with extensive high or focal high SI on T2-weighted or STIR imaging; for the chronic stage, appearance of low SI on both T1-weighted imaging and T2-weighted or STIR image is found.
DISCUSSION
Osteomyelitis embraces a wide spectrum and shows different features, depending on the clinical stage. Treatment requires a multidisciplinary ap- proach. Acute osteomyelitis usually responds to an- timicrobial therapy, whereas chronic osteomyelitis usually requires surgical procedure such as curettage of the necrotic bone and granulation tissue, and se- questrectomy.13,14 Radical surgical procedure, such as decorticotomy, is effective in the treatment of diffuse sclerosing osteomyelitis of the jaw.15,16 The diagnostic criteria should ideally be discussed in relation to the treatment procedures and clinical out- comes. In the present study, however, no relationship was found between MRI findings and selection of treatment procedures. It will be necessary in the future to examine whether MRI image can contribute to the choice of therapeutic method.
Various imaging features have been reported accord- ing to the activity of inflammation.2-6,10,11,17
Magnetic resonance imaging examination would be effective to differentiate osteomyelitis from other diseases with similar clinical symptoms; MRI images of osteomyeli- tis have been characterized by changes in SI of the bone Table II. Relationship between MRI features and bony appearance in 32 patients in the chronic group
MRI features Bony appearances on conventional radiography and CT
T1-weighted image
T2-weighted or
STIR image Lytic Mixed Sclerotic Sequestration No change
Low Extensive high 7 4 6 2
Focal high 1 1 2 2
Low 1
No change
No change Extensive high
Focal high 4*
Low
No change 2*
Total 7 5 8 2 10
MRI, magnetic resonance imaging; CT, computed tomography; STIR, short TI inversion recovery.
*These 6 patients could not be diagnosed as osteomyelitis.
marrow, such as decrease on T1-weighted images and increase on T2-weighted images.2-4,6,10 The authors emphasize the utility of MRI to depict the bone marrow changes in the acute stage before bony changes are visualized on conventional radiography. As for the chronic stage, Schuknecht et al.3have reported low SI both on T1- and T2-weighted images in some of the subjects. Their conventional images demonstrate exten- sive sclerosis of the cancellous bone. These imaging features, which were also observed in one of the present cases, can be a characteristic feature in the chronic stage. The STIR image is a recently introduced se- quence.18 This technique enables us to evaluate T2 signals more easily with fat tissue suppression.9,10Ad- ministration of gadolinium-DTPA may add very impor- tant information, such as a noncalcified periosteal re-
action, definition of the limit of the sequestrum, and extension of the inflammation to soft tissue.3 These features were also detectable on STIR image, and there- fore sequence with gadolinium-DTPA was not always used for making of MRI diagnostic criteria.
Although various MRI features have been clarified, no practical criteria that can be applied to all clinical stages have been proposed for the diagnosis of man- dibular osteomyelitis. This may be attributable to the lack of studies, which include these subjects without osteomyelitis but who were initially suspected as hav- ing osteomyelitis. Almost all of the previous studies investigated only those subjects with a definitive diag- nosis of osteomyelitis. So features that can differentiate osteomyelitis from other diseases showing similar clin- ical features have not been reported. The finding clar- Fig. 5. 39-year-old woman. She felt spontaneous pain and gingival swelling in the molar region of the left mandible for 4 months.
She underwent extraction of the second and third molars of the left mandible 1 month previously, but her subjective symptoms persisted. Her trismus increased gradually. A, Panoramic tomography showed sclerotic change in the area limited around the socket of the extracted teeth (arrow). Magnetic resonance imaging examination was performed 1 week later. B, The T1-weighted image showed a low SI area in the premolar to molar region of the left mandible (arrow). C, The STIR image showed a focal high SI area in the molar region (arrow).
ified in the present study, focal high SI on T2-weighted or STIR image without definitive changes on T1- weighted image, can be a significant feature that indi- cates a localized inflammation, such as marginal peri- odontitis, that should not be diagnosed as osteomyelitis.
The criteria proposed here support almost all of the traditional criteria that have been reported previously.1,2,7,8,17,19-21
However, an effective use of these criteria requires 2 sequences, T1-weighted and T2-weighted or STIR, in imaging. At present, MRI with 2 sequences is the first-choice examination when osteomyelitis is clinically suspected. However, ad- vanced techniques such as diffusion-weighted MRI are constantly being developed for MRI and should be examined as they come available.
For evaluation of response to treatment and long- term follow-up, another protocol may be required. In such an instance, MRI examination with administration of gadolinium-DTPA5and FDG-labeled positron emis- sion tomography may be effective.22-24
In conclusion, when osteomyelitis is suspected, es- pecially for subjects without positive features of osteo- myelitis on CT and conventional images, MRI with the 2 sequences used in this study should be performed.
REFERENCES
1. Jacobsson S. Diffuse sclerosing osteomyelitis of the mandible.
Int J Oral Surg 1984;13:363-85.
2. Schuknecht B, Valavanis A. Osteomyelitis of the mandible.
Neuroimaging Clin N Am 2003;13:605-18.
Fig. 6. 36-year-old man. He felt dull pain in the molar region of the left mandible following removal of a full crown of the second molar 3 months previously. A, Panoramic tomography showed diffusely sclerotic change in the left mandibular molar region (arrows). The range extended from the mandibular notch of the ramus to the lower margin of the mandible. Magnetic resonance imaging examination was performed 1 week later. Both T1-weighted image (B) and STIR image (C) showed low SI in the widespread area of the ramus (arrows).
3. Schuknecht BF, Carls FR, Valavanis A, Sailer HF. Mandibular osteomyelitis: evaluation and staging in 18 subjects, using mag- netic resonance imaging, computed tomography and conven- tional radiographs. J Craniomaxillofac Surg 1997;25:24-33.
4. Kaneda T, Minami M, Ozawa K, Akimoto Y, Utsunomiya T, Yamamoto H, et al. Magnetic resonance imaging of osteomyeli- tis in the mandible. Comparative study with other radiologic modalities. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995;79:634-40.
5. Bachmann G, Rossler R, Klett R, Rau WS, Bauer R. The role of magnetic resonance imaging and scintigraphy in the diagnosis of pathologic changes of the mandible after radiation therapy. Int J Oral Maxillofac Surg 1996;25:189-95.
6. Seabold JE, Simonson TM, Weber PC, Thompson BH, Harris KG, Rezai K, et al. Cranial osteomyelitis: diagnosis and fol- low-up with In-111 white blood cell and Tc-99m methylene diphosphonate bone SPECT, CT, and MR imaging. Radiology 1995;196:779-88.
7. Yoshiura K, Hijiya T, Ariji E, Sa do B, Nakayama E, Higuchi Y, et al. Radiographic patterns of osteomyelitis in the mandible.
Plain film/CT correlation. Oral Surg Oral Med Oral Pathol 1994;78:116-24.
8. Ida M, Tetsumura A, Kurabayashi T, Sasaki T. Periosteal new bone formation in the jaws. A computed tomographic study.
Dentomaxillofac Radiol 1997;26:169-76.
9. Zanetti M, Bruder E, Romero J, Hodler J. Bone marrow edema pattern in osteoarthritic knees: correlation between MR imaging and histologic findings. Radiology 2000;215:835-40.
10. Lee K, Kaneda T, Mori S, Minami M, Motohashi J, Yamashiro M. Magnetic resonance imaging of normal and osteomyelitis in the mandible: assessment of short inversion time inversion re- covery sequence. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2003;96:499-507.
11. Reinert S, Widlitzek H, Venderink DJ. The value of magnetic resonance imaging in the diagnosis of mandibular osteomyelitis.
Br J Oral Maxillofac Surg 1999;37:459-63.
12. Termaat MF, Raijmakers PG, Scholten HJ, Bakker FC, Patka P, Haarman HJ. The accuracy of diagnostic imaging for the assess- ment of chronic osteomyelitis: a systematic review and meta- analysis. J Bone Joint Surg Am 2005;87:2464-71.
13. Kim SG, Jang HS. Treatment of chronic osteomyelitis in Korea.
Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001;92:394-8.
14. Merkesteyn JP, Groot RH, van den Akker HP, Bakker DJ, Borgmeijer-Hoelen AM. Treatment of chronic suppurative os- teomyelitis of the mandible. Int J Oral Maxillofac Surg 1997;26:450-4.
15. Montonen M, Iizuka T, Hallikainen D, Lindqvist C. Decortica- tion in the treatment of diffuse sclerosing osteomyelitis of the mandible. Retrospective analysis of 41 cases between 1969 and 1990. Oral Surg Oral Med Oral Pathol 1993;75:5-11.
16. Kuroiwa Y, Matsuura H, Abe A, Kato M, Ariji Y, Kurita K.
Clinical observations of mandibular chronic osteomyelitis: com- bination therapy with decortication and intra-arterial infusion of antibiotics. Jpn J Oral Maxillofac Surg 2006;52:322-5 (in Japanese).
17. Flygare L, Norderyd J, Kubista J, Ohlsson J, Vallo-Christiansen J, Magnusson B. Chronic recurrent multifocal osteomyelitis in- volving both jaws: report of a case including magnetic resonance correlation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1997;83:300-5.
18. Bydder GM, Young IR. MR imaging: clinical use of the inver- sion recovery sequence. J Comput Assist Tomogr 1985;9:
659-75.
19. Orpe EC, Lee L, Pharoah MJ. A radiological analysis of chronic sclerosing osteomyelitis of the mandible. Dentomaxillofac Ra- diol 1996;25:125-9.
20. Suei Y, Taguchi A, Tanimoto K. Radiographic evaluation of possible etiology of diffuse sclerosing osteomyelitis of the man- dible. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1997;84:571-7.
21. Suei Y, Taguchi A, Tanimoto K. Diagnosis and classification of mandibular osteomyelitis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2005;100:207-14.
22. Hino S, Murase R, Terakado N, Shintani S, Hamakawa H.
Response of diffuse sclerosing osteomyelitis of the mandible to alendronate: follow-up study by 99mTc scintigraphy. Int J Oral Maxillofac Surg 2005;34:576-8.
23. Hakim SG, Bruecker CW, Jacobsen HCh, Hermes D, Lauer I, Eckerle S, et al. The value of FDG-PET and bone scintigraphy with SPECT in the primary diagnosis and follow-up of subjects with chronic osteomyelitis of the mandible. Int J Oral Maxillofac Surg 2006;35:809-16.
24. Ida M, Sakurai J, Tetsumura A, Kurabayashi T, Omura K, Amagasa T. MRI findings of osteomyelitis of the mandible.
Dental Radiol 2005;45:53-8 (in Japanese).
Reprint requests:
Yoshiko Ariji, DDS, PhD
Department of Oral and Maxillofacial Radiology Aichi-Gakuin University School of Dentistry 2-11 Suemori-dori, Chikusa-ku
Nagoya, 464-8651, Japan yoshiko@dpc.agu.ac.jp
![TraditionalMLCalgorithmsmainlytacklethebatchMLCproblem,wheretheinputdataarepresentedinabatch[24,28].Nevertheless,inmanyMLCapplicationssuchase-mailcategorization[22],multi-labelexamplesarriveasastream.Onlineanalysisistherefore dimensionreducermotivatedbyma](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACH5BAEAAAAALAAAAAABAAEAAAICRAEAOw==)