Manuscript title: The effect of three dimensional printing hinged ankle foot orthosis in chronic stroke patients

Background Three-dimensional printing (3DP) is a promising technique utilized in orthosis fabrication, including ankle foot orthosis for stroke patients. However, the effects on ankle biomechanics remains unclear. To compare the plantar pressure distribution and patient’s subjective experience in chronic stroke patients during 3DP hinged ankle foot orthosis (3DP-HAFO) and anterior ankle foot orthosis (A-AFO) walking Methods Ten patients with rst-ever unilateral stroke were enrolled in this study. All patients performed 10-meter walk test in 3 different conditions, including 3DP-HAFO walking, A-AFO walking, and bare foot walking. The plantar pressure parameters including contact area, maximum force, and peak pressure were collected using Pedar X insole system. Gait asymmetry analysis of the plantar pressure parameters was conducted. Quebec User Evaluation of Satisfaction with Assistive Technology (QUEST) was used for patient’s satisfaction. analysis illustrated more even medial midfoot contact area during 3DP-HAFO walking compared to bare foot walking while A-AFO walking In QUEST survey, 3DP-HAFO in tting and 3DP-HAFO affected limb during single stance phase. Posterior leaf AFO design is as useful as A-AFO in balance control for stroke patients. Both types of AFO exert effects of decreased excessive ankle plantar exions during swing and stance phase 15 . Adding hinge design on posterior leaf AFO can provide additional plantarexion control by lever arm and reduce initial toe contact of stance phase in hemiplegic patients 16 . The hinged AFO rendered the greatest support of dorsiexion during stance compared to posterior leaf and solid AFO 17 . Thus, both A-AFO and posterior design AFO can improve equinus foot pattern during walking. When it comes to inversion deformity, research revealed that A-AFO decreased ankle inversion during both stance and swing phases, while posterior AFO can only correct ankle inversion during swing phase 18 . Previous studies also support this perspective that A-AFO can enhance ankle medial-lateral control and improve ankle stability 6, 19 . Our nding illustrated that 3DP-HAFO can increase medial midfoot weight bearing compared to bare foot walking, and the effect was more signicant in contrast to A-AFO walking, reecting on decreased lateral midfoot peak pressure and increased medial midfoot peak pressure. The current evidence suggested that 3DP-HAFO can not only decrease ankle equinus pattern but also improve varus deformity. This nding has never been proposed in previous studies with posterior AFO without 3DP technique. The orthosis made of 3DP technique was more t, which may provide better ankle medial-lateral control. Further investigation is necessary to validate the current ndings.


Introduction
Stroke is the second leading cause of death worldwide, and the sequela of stroke is one of the main causes of adult disability, with up to 50% of stroke survivors being chronically disabled 1 . Post-stroke hemiplegia remains a long-term complication of stroke, which limits the patient's physical performance and mobility in daily life 2 . The restoration of functional ambulation is one of the priorities while setting rehabilitation goals in post-stroke patients. Ankle foot orthosis (AFO) is an orthotic device commonly fabricated to provide ankle stability during stance phase of gait cycle, and facilitate foot clearance during swing phase 3,4 .
Patients with stroke often wear the anterior leaf type AFO, called anterior AFO (A-AFO), for ankle support during ambulation after their condition become stabilized 5 . These orthoses improve the stability of ankle joints through drop foot reduction and the ability of lateral weight shifting through mediolateral support 6 . While A-AFO has above advantages, there are some pitfalls wearing A-AFO. Since the A-AFO is designed to protect and immobilize affected ankle joints, dynamic exion is limited. Moreover, there are no dynamic hinges designed in A-AFO, the range of ankle dorsi exion and plantar exion is restricted. The ankle support may not be enough because of lack of coverage at the heel. In addition, A-AFO is handmade, which let the contour and the size be slightly different from each product. This probably will disturb patients once changing new A-AFO.
The three-dimensional printing (3DP) technique, one of the most recent computer aided manufacturing techniques, has been introduced with emerging approaches to fabricate components of custom foot orthosis. Orthoses made with 3DP technique have several advantages, including easy production without the need of delicate skill compared with hand-made orthoses, easy reproduction with consistent quality once personalized 3DP modeling le was built and mass customized. The dimensional accuracy and manufacturing precision of 3DP technique has been validated 7 . Recent study illustrated that 3DP orthosis has positive subjective comfort rating 8 . Use of 3DP ankle foot orthosis (3DP-AFO) has shown at least equivalent performance to the handcrafted posterior leaf AFO 9 . However, the effect of 3DP-AFO compared with A-AFO is not reported. Also, it is unclear whether 3DP-AFO could affect plantar pressure distribution in stroke population.
In this study, we fabricated a 3DP-AFO with hinge and posterior leaf design (3DP-HAFO) to compare plantar pressure distribution and gait asymmetry among stroke patients wearing 3DP-HAFO, A-AFO or bare foot walking.

Participants
Participants were enrolled with con rmed diagnosis of stroke from brain computed tomography or magnetic resonance imaging. The inclusion criteria were as following: 1) rst-ever unilateral stroke (hemorrhagic or ischemic), 2) 20 years or older, 3) at least 3 months after stroke, 4) Functional Ambulation Category score of 3 or more, and 5) Brunnstrom stage of lower extremity III-IV. Patients were excluded if they had presence of low extremities peripheral vascular disease, sequelae of previous neurologic or orthopedic disorder that could impair locomotion, joint contracture in the lower extremities, skin problems, severe cognitive or visuospatial dysfunction and/or severe medical illness. All participants were informed of the study and submitted a written informed consent.

Orthosis design and fabrication
Fabrication of three-dimensional printing hinged ankle-foot orthosis (3DP-HAFO) First, we used 3D scanning system (Sense2 3D scanner, 3D SYSTEMS) to scan the shape of the affected leg to acquire 3D modelling le with triangle mesh architecture ( Figure 1a). Then, orthosis software (Rhinoceros®, Robert McNeel & Associates) was loaded. Medial and lateral malleolus as anatomical landmarks were manually marked at the heel to perform standard positioning points and reference lines according to preprogrammed orthotic template design derived from anthropometric data of normal, healthy volunteers. Ankle joint axis was adjusted to a neutral position by eversion (Figure 1b). The AFO hinged joint component socket was designed and built for assembly after printing (Figure 1c). The designed AFO was printed using a fused lament fabrication (FFF) type 3D printer (MINGDER 3D Printing 500S). Poly lactic acid (MINGDER 3D Printing) was used as printing material (extruders temperature 155-170°C) (Figure 1d). After printing out the designed AFO, components were trimmed to smoothen surface and hinge joint was assembled (Figure 1e).

Fabrication of anterior ankle-foot orthosis (A-AFO)
Anterior AFO is an anterior leaf orthosis made of thermoplastic material (CMC medical devices, 3.2cm in thickness, 55 o C-75 o C in molding temperature). It is cropped and molded directly to the lower leg under sitting position, with knee exion 90 degrees and ankle dorsi exion 5 degrees. The pretibial pad was added to reduced friction. The sole was fabricated around metatarsal area just behind the metatarsal head.

Protocol
All participants walked on a 10-meter walkway under three conditions in a random order, including walking with 3DP-HAFO, walking with A-AFO, and walking without orthosis (bare foot walking). The 3DP-HAFO and A-AFO were applied to hemiplegic leg. All participants wore standard shoes during the tests. The pedar®-X (Novel GmbH, Munich, Germany) insoles were placed within each shoe beneath the sole. The participants completed four walking trials for each condition. Participants were timed as they walked at a comfortable self-selected speed along the walkway. To ensure consistency of walking speed, any trial was eliminated and repeated if the time differed by more than 5% of the original trial time. To familiarize with orthosis wearing, participants tried and adjusted between the 3DP-HAFO and A-AFO orthoses for at least one month, until they felt comfortable on the day before the trials.

Pressure analysis equipment
Plantar pressures were measured using the pedar®-X system (Novel GmbH, Munich, Germany), which has been demonstrated to be a valid and reliable in-shoe pressure measurement system as described in the literature 10,11 . The pedar®-X insole comprised of 99 capacitive sensors arranged in a grid and embedded within a thin exible insole. The sampling frequency was 50 Hz. The pressure insoles were zeroed as described by the manufacturer's guidelines prior to the rst walking trial of each condition. Measured plantar foot pressure data were transmitted by using a Bluetooth connection to a computer for recording.
To determine gait performance with different AFO types, the gait speed and cadence were recorded during 3DP-HAFO walking, A-AFO walking and bare foot walking. The gait speed was calculated as time spent on 10-meter walking. The cadence was calculated as number of steps in one minute.
The contact area, maximum force, and peak pressure of each section were recorded. Data were compared in 4 mask regions corresponding to anatomically relevant areas of the foot (Figure 2), namely forefoot(distal 40% of foot length), lateral midfoot, medial midfoot, and hindfoot(proximal 27% of foot length), based on each participant's anteroposterior foot length. To determine gait asymmetry, plantar parameters were calculated following the formula below, where V indicated values of contact area, maximum force, and peak pressure in each section 12 : To evaluate participant satisfaction wearing different types of AFO, the Quebec User Evaluation of Satisfaction with Assistive Technology (QUEST) was used 10 . Each participant lled out the questionnaire during the out-patient clinic follow up approximately 2 months after orthosis wearing.

Statistical analysis
Statistical analysis was performed using Prism 8.0 for Windows (GraphPad Software, San Diego, USA). All measurements were analyzed for the difference among 3DP-HAFO walking, A-AFO walking and bare foot walking using Friedman test. Dunn's multiple comparisons test was used for post hoc analysis. The difference was considered statistically signi cant at p<0.05.

Participants basic demographics
The current study enrolled total 10 hemiplegic stroke participants, including 8 men and 2 women. The mean age of enrolled participants was 54 years old. The mean duration since stroke onset at the time of enrollment was 13 months. Four participants had ischemic stroke, and 6 had hemorrhagic stroke. The Brunnstorm motor recovery stages of affected lower extremities were stage III and IV (stage III in 5 participants, and stage IV in 5 participants). The Functional Ambulation Category was 3.1 (range 3−4). The manual muscle test of the affected ankle dorsi exor ranged from poor grade to good grade. The Modi ed Ashworth scale of affected ankle joint ranged from 1~1+ (1 in ve participants, 1+ in 5 participants). All participants were able to walk independently without cane under supervision, although different degrees of ankle inversion were observed in all participants during walking.
Changes in affected limb contact area, maximum force and peak pressure among different AFO types The contact area of medial midfoot on the affected limb was signi cantly increased in both 3DP-HAFO (p=0.01) and A-AFO (p=0.04) walking compared to bare foot walking ( Table 1). The maximum force of lateral midfoot on the affected limb was increased in A-AFO walking compared to bare foot walking (p=0.04), while those of 3DP-HAFO walking did not show signi cant difference ( Table  2). In addition, the peak pressure at medial midfoot on the affected limb was signi cantly increased in 3DP-HAFO walking in comparison to bare foot walking (p=0.01), while the peak pressure at lateral midfoot on the affected limb showed signi cant increase in A-AFO walking in comparison to bare foot walking (p=0.02) ( Table 3).
Changes in unaffected limb contact area, maximum force and peak pressure among different AFO types The contact area of medial midfoot on the unaffected limb was signi cantly increased in 3DP-HAFO walking compared to A-AFO walking (p=0.01) and bare foot walking (p=0.04). The signi cant change in maximum force on the unaffected limb was observed only at the medial midfoot, with increased maximum force in 3DP-HAFO walking compared to bare foot walking (p=0.04). Also, the peak pressure was only observed to be signi cant different at the medial midfoot on the unaffected limb, with increased medial midfoot peak pressure in 3DP-HAFO walking compared to both A-AFO walking(p=0.04) and bare foot walking(p=0.01). The details of contact area, maximum force and peak pressure during three conditions of walking were shown in Table 1-3.

Gait asymmetry
In both 3DP-HAFO and A-AFO walking, signi cant change in the asymmetric index was observed only in the contact area of medial midfoot ( Table 4). The asymmetric index of medial midfoot contact area were signi cantly improved in 3DP-HAFO as compared to barefoot walking (p=0.04), while A-AFO walking also revealed similar trend but failed to reach statistical signi cance.

Changes in gait performance among different AFO types
The gait speed and cadence during 3DP-HAFO walking, A-AFO walking and bare foot walking were shown in Table 5. There was no signi cance difference in gait speed and cadence among the three groups.

Participant's satisfaction
The questionnaire survey showed that participants felt 3DP-HAFO outweighed A-AFO in aspects of dimensions, durability, comfort and effectiveness. However, weight and convenience of wearing were two major concerns when wearing 3DP-HAFO (Table 6).

Discussion
In this study, we investigated the gait performance and plantar pressure effects of hinged AFO with 3DP technique and automated designed by CAD software on hemiplegic stroke population. The results suggested plantar pressure increased at medial midfoot with improved contact area asymmetry in 3DP-HAFO walking.
For stroke patients, walking with insu cient ankle dorsi exion are common due to joint stiffness and spasticity 13 . Clinically, patients experienced forefoot drop with dynamic equinovarus deformity throughout all gait cycle 14 . The abnormal gait pattern in stroke patients result in loss of mid foot weight bearing in stance phase. In the current study, increased contact area and peak pressure at the medial midfoot was observed during 3DP-HAFO walking as compared to bare foot walking. Increased medial midfoot contact area was also found in A-AFO walking. However, peak pressure was not increased in medial foot area, but in lateral midfoot during A-AFO walking. These ndings indicated that weight bearing is more in lateral side than in medial side in A-AFO walking compared to 3DP-HAFO walking. The results proposed that wearing 3DP-HAFO can partially correct dynamic equinovarus deformity and increase affected limb weight bearing, which can improve balance of affected limb during single stance phase.
Posterior leaf AFO design is as useful as A-AFO in balance control for stroke patients. Both types of AFO exert effects of decreased excessive ankle plantar exions during swing and stance phase 15 . Adding hinge design on posterior leaf AFO can provide additional plantar exion control by lever arm and reduce initial toe contact of stance phase in hemiplegic patients 16 . The hinged AFO rendered the greatest support of dorsi exion during stance compared to posterior leaf and solid AFO 17 . Thus, both A-AFO and posterior design AFO can improve equinus foot pattern during walking. When it comes to inversion deformity, research revealed that A-AFO decreased ankle inversion during both stance and swing phases, while posterior AFO can only correct ankle inversion during swing phase 18 . Previous studies also support this perspective that A-AFO can enhance ankle medial-lateral control and improve ankle stability 6,19 . Our nding illustrated that 3DP-HAFO can increase medial midfoot weight bearing compared to bare foot walking, and the effect was more signi cant in contrast to A-AFO walking, re ecting on decreased lateral midfoot peak pressure and increased medial midfoot peak pressure. The current evidence suggested that 3DP-HAFO can not only decrease ankle equinus pattern but also improve varus deformity. This nding has never been proposed in previous studies with posterior AFO without 3DP technique. The orthosis made of 3DP technique was more t, which may provide better ankle medial-lateral control. Further investigation is necessary to validate the current ndings.
Furthermore, stroke patients are prone to bear more weight on their unaffected leg [20][21][22] . The asymmetry of the contact pressure and area was found in stroke patients due to decreased contact area, force and pressure of the affected side compared with unaffected side 23 . The imbalance of weight distribution during walking will increase the risk of falling 22 . Training approaches emphasizing symmetric walking showed improvement of balance and functional outcomes 24,25 . Symmetrical walking after stroke played a crucial role for functional restoration 26 . In our study, the asymmetry of contact area was improved in 3DP-HAFO walking, suggesting that 3DP-HAFO walking may improve the gait through more symmetric weight distribution in both lower limbs. In contrast to A-AFO, 3DP-HAFO has stronger mid shank extension and may strengthen ankle stability and correct the dynamic equinovarus deformity during walking. The posterior leaf design of 3DP-HAFO may provide better sensory stimulation on hemiplegic patients' feet as compared to A-AFO design, which is xed on foot arch 27-29 .
One of the outstanding characteristics of 3DP technique is individualization. In this study, 3DP-HAFO showed better satisfaction in comfort item in the QUEST compared to A-AFO. Although there is no difference in functional outcome of gait speed and cadence between 3DP-AFO and A-AFO walking, 3DP-HAFO outweighed A-AFO in items of effectiveness in the QUEST. The result implied that stability and balance, rather than gait speed, are the primary concerns to evaluate satisfaction of orthosis use in walking assist of hemiplegic stroke population. In addition, 3DP-HAFO was more durable than A-AFO. It may be attributed that 3DP-AFO was made from high temperature process, while A-AFO was made from low-temperature thermoplastic material. There are two drawbacks of our 3DP-HAFO. First, weight of 3DP-HAFO was heavier compared to A-AFO. This re ected on the QUEST. The other was the ease of using and adjustment. Due to posterior leaf design of 3DP-HAFO, participants needed more time to take on and off shoes when using 3DP-HAFO. Shoes selection was also restricted to at sole to t rigid orthotic bottom.
There are some limitations in the current study. As previous study showed that AFO has immediate effect in improving gait and balance 30 , temporal follow up is not analyzed in this study. Moreover, kinematic and kinetic assessments were not performed for gait analysis. Thus, further investigation with longitudinal follow up is necessary to validate the current ndings.
In conclusion, 3DP-HAFO may improve the ankle stability and gait symmetry, but not the gait speed. Further studies are warranted to determine whether long-term use of 3DP-HAFO and combined rehabilitation training are effective in improving the gait patterns and functional walking in hemiparetic stroke population.

Clinical Messages
Clinical Messages: 3DP is a feasible technique in ankle foot orthosis fabrication process.
Wearing 3DP-HAFO may improve ankle medial lateral control and gait asymmetry in hemiplegic patients.
3DP-HAFO is more t, comfortable, effective and durable compared to traditional A-AFO.

Declarations
Acknowledgements: The authors thank Yi-Pei Chen, the research nurse of Clinical Trial Center of Kaohsiung Medical University Hospital for dedicating her time and efforts for the study.
Author contributions: CM Fu, YJ Chen, and CH Chen contributed to the conception and design of the work; CM Fu, CF Li, and YH Hsiao contributed to acquisition and analysis of data; CM Fu, YJ Chen, CF Li, YH Hsiao, FZ Sheen, and CH Chen contributed to interpretation of data; CM Fu and YJ Chen contributed to drafting of the manuscript; CF Li, YH Hsiao, FZ Sheen, and CH Chen contributed to revision of the manuscript; all authors contributed to nal approval of the manuscript.
Competing interests: The authors declare that there is no con ict of interest. Methods were managed in accordance with the approved guidelines.
All patients and caregivers were provided with information about the procedures and purpose of this study and provided written informed consent before inclusion.

27
Sadeghi-Demneh, E. The effects of orthotics on the sensori-motor problems of the foot and ankle after stroke (2011).     Table 4. Asymmetric index of 3DP-HAFO, A-AFO and bare foot walking